| |
Last updated on November 15, 2020. This conference program is tentative and subject to change
Technical Program for Thursday November 19, 2020
|
ThA42 |
L-2 |
R1: Robotics, Control Systems & Theory |
Regular Session |
|
11:00-11:15, Paper ThA42.1 | |
Unmanned Ground Vehicle for Detection of Permissible Exposure to Crude Oil Fume |
|
Tan, Gerhard | Polytechnic University of the Philippines |
Anacan, Rommel | Technological Institute of the Philippines |
Valondo, Sheila Carmina | Technological Institute of the Philippines |
Barte, Neil | Technological Institute of the Philippines |
Zagada, Aldrich Jay | Technological Institute of the Philippines |
Tabagan, Ryan Carlo | Technological Institute of the Philippines |
Memije, Christine Joy | Technological Institute of the Philippines |
Galag, Raphaelo | Technological Institute of the Philippines |
Cartin, Romulo | Technological Institute of the Philippines |
Nagallo, Jose Nicolas | Technological Institute of the Philippines |
Keywords: Robotics, Control Systems & Theory
Abstract: Crude oil is
extremely flammable & can cause eye, skin, gastrointestinal, and
respiratory irritation. Breathing the fumes from crude oil are known to
cause chemical pneumonia, irritation of the nose, throat, & lungs,
etc.. In the shipyard, workers are exposed to the fumes of crude oil
while doing maintenance on ships. Their safety & health can be at
risk while doing their job. This research aims to develop a device that
will detect & assess specific types of gases in order to determine
the permissible and impermissible time of work for employees' safety
precaution. Unmanned ground vehicle is the machine that operates while
in contact with the ground and without an on-board human presence. By
gathering the data from the sensors who has a specific sensitivity on
the substances in crude oil fumes, the data will then convert to a gas
concentration in a unit of ppm where distinct values can be analyzed.
Using data sets and Gradient Boost Algorithm, the system can train data
based on the output of the acquired signal which eventually will be more
accurate in determining the permissible exposure limit of a worker to
the crude oil fume. Results shows Gradient Boost Algorithm has an
average accuracy of 93.75% with a failure rate of 0.0823. The time
response of the design is 2.98 seconds. It is recommended that the
device used for getting the input gas sensor should be accurate to the
gas decomposition of the oil / petroleum you are trying to test.
|
|
11:15-11:30, Paper ThA42.2 | |
A Two-Step Adaptive Strategy for Simple Time-Varying Systems |
|
Subramanian, Karpagavalli | PES University |
George, Koshy | PES University |
Keywords: Robotics, Control Systems & Theory
Abstract: In recent
years, the need for controlling an unknown plant arises in various
fields like aircraft, automotive, and manufacturing industries in order
to provide quicker and more accurate response in the presence of
parametric uncertainties. Classical adaptive control may not as yet be
able to give the desired results when the plant is unknown with
time-varying parameters. In such a scenario, it is quite likely that the
response of the plant grow unbounded. Moreover, adaptive control leads
to better transient response only when the estimates of the plant
parameters are reasonably close to that of the actual values. In this
paper, we propose a two-step adaptive strategy that helps to quickly
converge into that compact region wherein the plant parameters reside.
Further, we demonstrate that our proposed method provides partial
answers to the adaptive control of linear time-varying plants.
|
|
11:30-11:45, Paper ThA42.3 | |
Suction-Port Position Control for Vacuum Working Robot to Improve Working Efficiency |
|
Tanaka, Kosuke | Mie Univesity |
Morimoto, Tatsuhiro | Mie University |
Yano, Ken'ichi | Mie University |
Arima, Toshihiko | Shinagawa Furnace |
Fukui, Shigeru | Shinagawa Furnace |
Keywords: Robotics, Control Systems & Theory
Abstract: When cleaning
the dirt pile around the conveyor with a vacuum car, there are problems
that workers collide with falling iron ore and suffer health problems by
breathing in dust. Therefore, it is necessary to develop a robot that
performs the cleaning with a vacuum car instead of workers. Many studies
on robots that perform cleaning work have discussed about automation
and cleaning efficiency such as automatic mapping and coverage path
planning. However, there is no discussion about cleaning efficiency and
automation of robots that can perform the cleaning with a vacuum car. To
perform the cleaning efficiently, it is necessary to avoid cleaning the
same place several times and leaving uncleaned areas because it
increases the cleaning time. Therefore, in this study, we first
formulated the area where the hose sucks the dirt pile. Then, we
proposed a method to derive the optimal path that minimizes the cleaning
time. Furthermore, we developed a control system that realizes the
optimal path. Finally, we conducted a verification experiment assuming
an aisle next to the conveyor. In the experiment, it was confirmed that
the coincidence rate with the optimal path was about 70% by the
suction-port position control system.
|
|
11:45-12:00, Paper ThA42.4 | |
Stabilization of a Quadrotor with Safety Harness During Collision and Hover |
|
Aoki, Tomoya | Kyoto Institute of Technology |
Higashi, Yoshiyuki | Kyoto Institute of Technology |
Keywords: Robotics, Control Systems & Theory
Abstract: This research
proposes a controller that solves the problems of divergence of
behaviors during a collision and steady-state deviation during hovering.
An attitude controller that sets the target attitude angle to 0 is used
to stabilize the behaviour during a collision. And a designed
controller that treat a quadrotor with a safety harness as a double
pendulum solves steady-state errors during hover by calculating inputs
that reduce the energy of the system. We tested the proposed controller
by using MATLAB and virtual robot experimentation platform (V-REP). The
results of the simulation showed that a quadrotor with a safety harness
prevents post-collision divergence and allows for hovering directly
below the rod.
|
|
12:00-12:15, Paper ThA42.5 | |
Simulation of a Reconfigurable Spherical Robot IV for Confined Environment |
|
Laksanacharoen, Pudit | King Mongkut's University of Technology North Bangkok |
Bunathuek, Natthaphon | KMUTNB |
Keywords: Robotics, Control Systems & Theory
Abstract: This work
presents a new type of Reconfigurable Spherical Robot for confined
environment. The robot is a spherical shape with three legs kept inside
spherical shell. Each leg has four degrees of freedom. All three legs
can be extended for two types of locomotion such as legged locomotion
and rolling sphere. The simulation results performed in MATLAB Simulink
have shown that the robot can move in a confined environment by rolling
its spherical shape and can traverse over obstacles.
|
|
ThA43 |
L-3 |
ML9: Machine Learning, Cloud and Data Analytics |
Regular Session |
Chair: Tokuda, Tomoki | ATR |
|
11:00-11:15, Paper ThA43.1 | |
Vision-Based Shrimp Feed Type Classification Using Fuzzy Logic |
|
Gamara, Rex Paolo | De La Salle University |
Bandala, Argel | De La Salle University |
Loresco, Pocholo James | De La Salle University |
Keywords: Machine Learning, Cloud and Data Analytics, Signal and Image Processing
Abstract: Shrimp farming
is a major industry covering 23% of Philippine annual aquaculture
production, which requires performing better management practices (BMPs)
including growth monitoring and feed management. Traditionally, growth
is monitored manually using analog weighing scale and caliper; but the
manual measurement is a tedious task for large-scale farming. Feed
management entails providing the most suitable feed type based on the
shrimp痴 current growth stage; furthermore, it addresses issues of
underfeeding and overfeeding. The limitations of manual method led to
the implementation of computer vision applications for growth
measurement. However, existing vision-based measurement studies are not
yet applied for feed management. This paper presented a fuzzy-logic
based shrimp feed type classification system utilizing Mamdani痴
methodology. The output classes are Starter, Grower, and Finisher based
on the three inputs: pixel area, length, and weight. The system was
developed using the FIS feature of the MATLAB Fuzzy Logic toolbox. The
classification system was evaluated and resulted to 93.33% correct
classification accuracy. Based on these results, it can be concluded
that fuzzy logic can be utilized to determine the suitable shrimp feed
type corresponding to the input features.
|
|
11:15-11:30, Paper ThA43.2 | |
Transfer Learning Approach for the Classification of Conidial Fungi (Genus Aspergillus) Thru Pre Trained Deep Learning Models |
|
Mital, Matt Ervin | De La Salle University |
Tobias, Rogelio Ruzcko | De La Salle University |
Villaruel, Herbert | De La Salle University |
Maningo, Jose Martin | De La Salle University |
Billones, Robert Kerwin | De La Salle University |
Vicerra, Ryan Rhay | De La Salle University |
Bandala, Argel | De La Salle University |
Dadios, Elmer | De La Salle University |
Keywords: Machine Learning, Cloud and Data Analytics, Signal and Image Processing
Abstract: The Aspergillus
genus is deemed relevant for distinction and classification in the
field of food, agriculture and medicine. As there are harmful and useful
ones, it adds to the necessity of correct classification.
Categorization of this conidial fungi is usually done through manual
microscopical procedures which apparently has a degree of
subjectiveness. In order to classify Aspergillus samples faster and more
accurately, technology, specifically image processing and machine
learning are incorporated in this study. Pre-trained deep learning
models are employed in classifying 9 kinds of Aspergillus. The
methodology is generally comprised of pre-processing, deep-learning
(training) and performance evaluation. Performance evaluation pertains
to the validation accuracy and running times of the system after
training through visual display of graphs and tabulation of acquired
data. This study achieved a 93.3333% testing accuracy proving that the
transferred knowledge is accurate, compatible and reliable.
|
|
11:30-11:45, Paper ThA43.3 | |
Hybrid Tree-Fuzzy Logic for Aquaponic Lettuce Growth Stage Classification Based on Canopy Texture Descriptors |
|
Tobias, Rogelio Ruzcko | De La Salle University; Asia Pacific College |
Mital, Matt Ervin | De La Salle University |
Concepcion, Ronnie II | De La Salle University |
Lauguico, Sandy | De La Salle University |
Alejandrino, Jonnel | De La Salle University |
Montante, Samboy Jim | Asia Pacific College |
Vicerra, Ryan Rhay | De La Salle University |
Bandala, Argel | De La Salle University |
Sybingco, Edwin | De La Salle University |
Dadios, Elmer | De La Salle University |
Keywords: Machine Learning, Cloud and Data Analytics, Signal and Image Processing
Abstract: Lettuce is one
of the most popular crops for urban farming because it is easy to grow
and it has high nutritional value. Moreover, it is adaptable and can be
combined with other food options, or it can be eaten alone without too
much preparation. Predicting lettuce growth can be crucial to find the
optimum maturity and harvest time. This paper proposed to use a model of
a hybrid tree-fuzzy logic approach, the classification tree was used to
select the most significant features from the texture features then the
fuzzy inference system was utilized in predicting the lettuce growth
stage classification. The hybrid system produced accurate results with
low percentage error and correct classifications. Based on these
results, the most accurate prediction can be observed in the head
development growth stage; the harvest growth stage has a slight
variance, while the vegetative stage has the most variance. Overall, the
trained hybrid system is reliable in predicting and identifying lettuce
growth stage classification.
|
|
ThA44 |
L-4 |
SIP5: Signal and Image Processing |
Regular Session |
Chair: Jia, Haohui | Nara Institute of Science and Technology |
|
11:00-11:15, Paper ThA44.1 | |
Predicting Valence and Arousal by Aggregating Acoustic Features for Acoustic-Linguistic Information Fusion |
|
Atmaja, Bagus Tris | JAIST |
Akagi, Masato | JAIST |
Hamada, Yasuhiro | JAIST |
Keywords: Signal and Image Processing, Multimedia Engineering, Disasters and Humanitarian Technology
Abstract: This paper
presents an evaluation of acoustic feature aggregation and
acoustic-linguistic features combination for valence and arousal
prediction within a speech. First, acoustic features were aggregated
from chunk-based processing for story- based processing. We evaluated
mean and maximum aggregation methods for those acoustic features and
compared the results with the baseline, which used majority voting
aggregation. Second, the extracted acoustic features are combined with
linguistic features for predicting valence and arousal categories: low,
medium, or high. The unimodal result using acoustic features aggregation
showed an improvement over the baseline majority voting on development
partition for the same acoustic feature set. The bimodal results (by
combining acoustic and linguistic information at the feature level)
improved both development and test scores over the official baseline.
This combination of acoustic-linguistic information targeted
speech-based applications where acoustic and linguistic features can be
extracted from the sole speech modality.
|
|
11:15-11:30, Paper ThA44.2 | |
An OCV Estimation Algorithm for Lithium-Ion Battery Using Kalman Filter |
|
Lin, Lei | Ritsumeikan University |
Fukui, Masahiro | Ritsumeikan University |
Keywords: Signal and Image Processing, Power & Energy
Abstract: This paper
discusses a method of open circuit voltage (OCV) estimation system for
lithium-ion batteries based on the join method of Kalman filter and RLS
method. We set mean filter to improve the stability of RC parameter
estimation. As result, the propose method performs stable and accurate
estimation in each test pattern.
|
|
11:30-11:45, Paper ThA44.3 | |
Semantic Segmentation of Femur Bone from MRI Images of Patients with Hematologic Malignancies |
|
Peng, Boyuan | The University of Aizu |
Ikeda, Shouhei | Aizu Medical Center |
Guo, Zhe | The University of Aizu |
Tsunoda, Saburo | Aizu Medical Center Fukushima Medical University |
Zhu, Xin | The University of Aizu |
Keywords: Biomedical Engineering, Signal and Image Processing, Machine Learning, Cloud and Data Analytics
Abstract: Femur bone
marrow MRI images may provide more information for better characterizing
hematologic malignancies and understanding their prognostics compared
with blind bone marrow biopsies and aspirates. However, the
interpretation of femur bone marrow MRI images is time-consuming and
intervariable among different physicians. To develop a computer-aided
diagnosis system for hematologic malignancies using femur bone marrow
MRI images, we propose a fully automatic method for femur bone
segmentation using deep learning. Five classical pretrained networks
such as U-Net and Segnet based on a backbone of Resnet50 network are
used for femoral bone segmentation. We conducted an experimental study
on 200 cases with hematologic malignancies. Through training using 149
patientsT1-enhanced MRI images, we obtained an average Dice
coefficient of 0.907 using Segnet based on a validation using 38
patients' T1-enhanced MRI images.
|
|
11:45-12:00, Paper ThA44.4 | |
Classification Method with CNN Features and SVM for Computer-Aided Diagnosis System in Colorectal Magnified NBI Endoscopy |
|
Odagawa, Masayuki | Cadence Design Systems |
Koide, Tetsushi | Hiroshima University |
Okamoto, Takumi | Cadence Design Systems |
Tamaki, Toru | Hiroshima University |
Raytchev, Bisser | Hiroshima University |
Kaneda, Kazufumi | Hiroshima University |
Yoshida, Shigeto | Medical Corporation JR Hiroshima Hospital |
Mieno, Hiroshi | Medical Corporation JR Hiroshima Hospital |
Tanaka, Shinji | Hiroshima University Hospital |
Keywords: Biomedical Engineering, Signal and Image Processing, Machine Learning, Cloud and Data Analytics
Abstract: This paper
presents a classification method for a Computer-Aided Diagnosis (CAD)
system in a colorectal magnified Narrow Band Imaging (NBI) endoscopy.
For the classification of a histologic findings, we consider an output
result of a lesion endoscopic image from a pre-learned Convolutional
Neural Network (CNN) as a feature vector and construct a set of Support
Vector Machines (SVMs) by learning a set of the CNN feature vectors. In
the video images, each frame has appearance changes such as blur, color
shift, reflection of light and so on and it affects classification
results. To improve the robustness of CAD system, we construct the SVM
learned by multiple image sizes data sets so as to adapt to the noise
peculiar to the video image. We confirmed that the proposed method
achieves higher robustness, stable, and high classification accuracy in
the endoscopic video image. The proposed method also can cope with
differences in resolution by old and new endoscopes and perform stably
with respect to the input endoscopic video image. We evaluated the
proposed method on a customizable embedded DSP core implemented into a
FPGA based prototyping system.
|
|
12:00-12:15, Paper ThA44.5 | |
Self Attentive Normalization for Automated Gleason Grading System |
|
Shin, Hong-Kyu | Korea University |
Hong, Sung-Hoo | Seoul St. Mary痴 Hospital |
Choi, Yeong-Jin | Seoul St. Mary痴 Hospital |
Shin, Yong-Goo | Korea University |
Park, Seung | Korea University |
Ko, Sung-Jea | Korea University |
Keywords: Biomedical Engineering, Signal and Image Processing, Machine Learning, Cloud and Data Analytics
Abstract: Recently,
convolutional neural networks (CNNs)-based automated Gleason grading
system for prostate cancer has been widely researched. However, these
systems still need further improvement to achieve pathologist-level
performance. To this end, this paper introduces a novel self-attentive
normalization (SAN) which is the first work to employ the attention
mechanism for the automated Gleason grading system. Unlike conventional
normalization techniques, e.g. batch normalization and instance
normalization, which learn a single affine transformation, the proposed
method can learn the element-wise affine transformation to focus on more
informative regions of the feature map. Since SAN requires a small
number of extra learning parameters, it can be integrated into existing
automated Gleason grading systems image classification seamlessly with
negligible overheads. Extensive quantitative evaluations show that, by
applying SAN to various CNN architectures, the diagnostic accuracy can
be significantly improved. For instance, we raise VGG-16's diagnostic
accuracy from 73.99% to 79.16% on the Harvard Dataverse.
|
|
12:15-12:30, Paper ThA44.6 | |
Anomaly Detection in Panoramic Dental X-Rays Using a Hybrid Deep Learning and Machine Learning Approach |
|
Verma, Dhruv | Hewlett Packard Enterprise |
Puri, Sunaina | Manipal College of Dental Sciences |
Prabhu, Srikanth | Manipal Institute of Technology |
Smriti, Komal | Manipal College of Dental Sciences |
Keywords: Signal and Image Processing, Biomedical Engineering, Machine Learning, Cloud and Data Analytics
Abstract: Automated
anomaly detection in panoramic dental x-rays is a crucial step in
streamlining post diagnosis treatment. It can reduce clinical time for a
patient and also aid in giving them faster access to medical care. In
this paper, we propose a hybrid deep learning and machine learning based
approach to detect evident dental caries/periapical infection, altered
periodontal bone height, and third molar impactions using panoramic
dental radiographs. We use a Convolutional Neural Network as a feature
extractor for an input image and use a Support Vector Machine to
classify the image as either "Normal" or "Anomalous" based on the
extracted features. We compare the performance of this model with the
performance of a Convolutional Neural Network and a Support Vector
Machine for the same classification task. We also compare our best model
with other existing models trained to detect carries and periodontal
bone loss. The results obtained with the hybrid deep learning and
machine learning approach outperformed the existing methods in the
literature.
|
|
ThA45 |
L-5 |
ME1: Multimedia Engineering |
Regular Session |
Chair: Chen, Na | Nara Institute of Science and Technology |
|
11:00-11:15, Paper ThA45.1 | |
Making Kids Learning Joyful Using Artistic Style Transferred YouTube VCs |
|
Kumar, Shobhan | Indian Institute of Information Technology, Dharwad |
Chauhan, Arun | Indian Institute of Information Technology, Dharwad |
Keywords: Engineering Education, Multimedia Engineering
Abstract: The rise of
E-learning systems offers a large number of open online educational
videos through portals like YouTube, MOOCs, which result in an enormous
volume of data for every inquisitive e-learner. This work proposes a
novel approach for recommending educational video clips (VCs) for
children's joyful learning. It caters to the need for hard-pressed
parents, who are unable to tutor their children efficaciously. As
children are unable to choose the recommended VCs on their own, the
proposed system offers a precise summary of the recommended VCs for
parents to effectively select the learning content for their children.
For joyful learning experience, the selected VCs are then style
transferred using our deep style transfer network. The extensive
evaluation methods demonstrate the effectiveness and practicability of
the proposed approach.
|
|
11:15-11:30, Paper ThA45.2 | |
Multi-Task Learning for Detection, Recovery, and Separation of Polyphonic Music |
|
Tan, Vanessa | University of the Philippines Diliman |
de Leon, Franz | University of the Philippines Diliman |
Keywords: Machine Learning, Cloud and Data Analytics, Signal and Image Processing, Multimedia Engineering
Abstract: Music
separation aims to extract the signals of individual sources from a
given audio mixture. Recent studies explored the use of deep learning
algorithms for this problem. Although these algorithms have proven to
have good performance, they are inefficient as they need to learn an
independent model for each sound source. In this study, we demonstrate a
multi-task learning system for music separation, detection, and
recovery. The proposed system separates polyphonic music into four sound
sources using a single model. It also detects the presence of a source
in the given mixture. Lastly, it reconstructs the input mixture to help
the network further learn the audio representation. Our novel approach
exploits the shared information in each task, thus, improving the
separation performance of the system. It was determined that the best
configuration for the multi-task learning is to separate the sources
first, followed by parallel modules for classification and recovery.
Quantitative and qualitative results show that the performance of our
system is comparable to baselines for separation and classification.
|
|
11:30-11:45, Paper ThA45.3 | |
Improved Viseme Recognition Using Generative Adversarial Networks |
|
Shreekumar, Jayanth | PES University |
P N, Vijay | PES University |
Shet, Ganesh Krishnamurthy | PES University |
S J, Preethi | PES University |
Krupa, Niranjana | PES University |
Keywords: Machine Learning, Cloud and Data Analytics, Signal and Image Processing, Multimedia Engineering
Abstract: The
proliferation of convolutional neural networks (CNN) has resulted in
increased interest in the field of visual speech recognition (VSR).
However, while VSR for word-level and sentence-level classification has
received much of this attention, recognition of visemes has remained
relatively unexplored. This paper focuses on the visemic approach for
VSR as it can be used to build language-independent models. Our method
employs generative adversarial networks (GANs) to create synthetic
images that are used for data augmentation. VGG16 is used for
classification both before and after augmentation. The results obtained
prove that data augmentation using GANs is a viable technique for
improving the performance of VSR models. Augmenting the dataset with
images generated using the Progressive Growing Generative Adversarial
Network (PGGAN) model led to an average increase in test accuracy of
3.695% across speakers. An average increase in test accuracy of 2.59%
was achieved by augmenting the dataset using images generated by the
conditional Deep Convolutional Generative Adversarial Network (DCGAN)
model.
|
|
11:45-12:00, Paper ThA45.4 | |
Convolutional Neural Network-Based Criminal Detection |
|
Verma, Harsh | International Institute of Information Technology Naya Raipur |
Lotia, Siddharth | IIIT Naya Raipur |
Singh, Anurag | IIIT Naya Raipur |
Keywords: Signal and Image Processing, Machine Learning, Cloud and Data Analytics, Multimedia Engineering
Abstract: Various recent
advancements in deep learning models have greatly boosted the
performance of semantic pattern recognition using images. Various state
estimation of an individual like emotional state and other certain
character features or traits can be estimated from the facial images.
With this motivation, in this work we are attempting to infer criminal
tendency or (crime prediction/detection) from facial images by using the
learning capabilities of various deep learning architectures. More
precisely two type of deep learning models we have used in this study:
standard convolutional neural network(CNN) architecture and pre-trained
CNN architectures, namely VGG-16, VGG-19, and InceptionV3. We have done a
performance comparative analysis among these models for efficiently
capturing criminal traits from a human face. The efficacy of the above
deep learning models was evaluated on a public database, National
Institute of Standards and Technology (NIST). To avoid any
discrepancies, we have only used male images in this work. It was found
that VGG CNN models are best performing models, especially in a limited
data scenario producing the classification accuracy of 99.5% in
identifying criminal faces.
|
|
ThB41 |
L-1 |
ML11: Machine Learning, Cloud and Data Analytics |
Regular Session |
Chair: Higashino, Takeshi | Nara Institute of Science and Technology |
|
13:30-13:45, Paper ThB41.1 | |
One-Class Support Vector Machine for Data Streams |
|
Bhat, Srinidhi | Stony Brook University |
Singh, Sanjay | Manipal Institute of Technology |
Keywords: Machine Learning, Cloud and Data Analytics, Devices, Materials & Processing, Social Implications of Technology
Abstract: In various
information systems, application learning algorithms have to act in a
dynamic environment where the acquired data is in data streams. In
contrast to static data mining, processing streams introduce an array of
computational and algorithmic stipulations. With the continuous input
of data in data streams, one would like a mechanism that automatically
identifies unusual events in the time series. The topic has been in the
limelight as it has huge potential for real-time activities. To show the
algorithm's robustness, we have trained the classifier to multiple
activities and its success in identifying each activity. The paper
explores the possibility of using the One-Class Support Vector Machine
(OCSVM) for novelty detection in data streams.
|
|
13:45-14:00, Paper ThB41.2 | |
Early Detection of Forest Fire Using Deep Learning |
|
Medi, Rahul | International Institute of Information Technology (IIIT) Naya Ra |
Karnekanti, Shiva Saketh | International Institute of Information Technology (IIIT) Naya Ra |
Attili, Sanjeet | International Institute of Information Technology (IIIT) Naya Ra |
Nenavath, Srinivas Naik | International Institute of Information Technology (IIIT) Naya Ra |
Keywords: Machine Learning, Cloud and Data Analytics, Disasters and Humanitarian Technology, Signal and Image Processing
Abstract: Forest fires
have become a serious threat to mankind. Besides providing shelter and
protection to a large number of living beings, they have been a major
source of food, wood, and a great supply of other products. Since
ancient times forests have played an important role in social, economic,
and religious activities and have enriched human life in a variety of
ways both material and psychological. To protect our nature from these
rapidly rising forest fires, we need to be cautious enough of every
decision we take which could lead to a disastrous end, once and for all.
So for the early detection of forest fires, we propose an image
recognition method based on Convolutional Neural Networks (CNN). We have
fine-tuned the Resnet50 architecture and added a few convolutional
layers with ReLu as the activation functions, and a binary
classification output layer which showed a huge impact on the training
and test results when compared to the other SOTA methods like VGG16 AND
DenseNet121. We achieved a training set accuracy of 92.27% and 89.57%
test accuracy with a stochastic gradient descent optimizer and we have
avoided the underfit/overfitting on the model with the help of the
Stochastic Gradient Descent (SGD) algorithm.
|
|
14:00-14:15, Paper ThB41.3 | |
Forecasting PV Panel Output Using Prophet Time Series Machine Learning Model |
|
Shawon, Md. Mehedi Hasan | BRAC University |
Akter, Sumaiya | BRAC University |
Islam, Md. Kamrul | BRAC University |
Ahmed, Sabbir | BRAC University |
Rahman, Md. Mosaddequr | BRAC University |
Keywords: Machine Learning, Cloud and Data Analytics, Power & Energy
Abstract: Due to climate
change effects, the demand for renewable energy is growing immensely
around the world. Photovoltaic (PV) panels are widely popular as a vital
source of renewable energy all over the world as well as in Bangladesh.
However, besides solar irradiance, the panel output is greatly affected
by some of the weather parameters like temperature, humidity, wind,
etc. Reliable forecasting of PV panel output is essential for capacity
planning in advance to efficiently manage the energy distribution. This
paper presents a method to forecast the PV panel output energy using a
machine learning model, known as the Prophet Model used for a univariate
time series forecasting. For this study, the PV panel generated data
are collected from an outdoor experimental set-up throughout the full
winter season in Bangladesh. Based on the data, forecasting of
one-day-ahead PV panel short circuit current is done, and then the
estimation of PV panel output energy is made. The results show the
proposed forecasting method to be quite encouraging and reliable one
while providing a higher coefficient of determination value with an
average 0.9772 for one-day-ahead PV panel output energy forecasting.
|
|
14:15-14:30, Paper ThB41.4 | |
An RNN Approach for Lithium-Ion Battery Internal Impedance Estimation |
|
Harada, Yuusaku | Ritsumeikan University |
Lin, Lei | Ritsumeikan University |
Fukui, Masahiro | Ritsumeikan University |
Keywords: Machine Learning, Cloud and Data Analytics, Power & Energy
Abstract: In this paper,
we propose an internal impedance estimation for lithium-ion battery
using LSTM neural network method. Compare to the conventional RLS
method, the LSTM performs stable estimation with less variance.
|
|
14:30-14:45, Paper ThB41.5 | |
Implementation of an IoT-Based DC Nanogrid in an Offshore Fish Farm |
|
Quek, Yang Thee | Republic Polytechnic |
Keywords: Machine Learning, Cloud and Data Analytics, Power & Energy, Marine and Offshore Engineering
Abstract: Offshore fish
farms remain a vibrant and important production sector for seafood in
the aquaculture industry. Being offshore with no grid-based power
supply, the continuity and quality of the power supply to the fish farm
is crucial. This paper presents an implementation of an Internet of
Things (IoT)-based dc nanogrid in an offshore fish farm. The standalone
self-sustainable dc nanogrid primarily consists of photovoltaic panels
as the power supply and a battery bank as the energy storage. An added
feature of the system is a variety of dc voltages with differentiated
sockets to cater to different dc-powered loads and appliances. There are
various sensor points in the system to acquire readings of temperature,
voltage and current for monitoring of the performance of the system.
The system is enhanced with IoT features, which give the user the option
to access the local computer remotely to monitor and manage the system
or to view real-time data on a cloud platform. Smart features are also
added in the system to detect abnormal readings of data and send warning
emails to the user.
|
|
14:45-15:00, Paper ThB41.6 | |
Genetic Algorithm-Based Dark Channel Prior Parameters Selection for Single Underwater Image Dehazing |
|
Almero, Vincent Jan | De La Salle University |
Concepcion, Ronnie II | De La Salle University |
Alejandrino, Jonnel | De La Salle University |
Bandala, Argel | De La Salle University |
Espala, Jason | De La Salle University |
Bedruz, Rhen Anjerome | De La Salle University |
Vicerra, Ryan Rhay | De La Salle University |
Dadios, Elmer | De La Salle University |
Keywords: Machine Learning, Cloud and Data Analytics, Signal and Image Processing
Abstract: Dehazing
through Dark Channel Prior (DCP), originally developed for land-based
images, has translated its potential for improving the quality of
underwater images. However, the DCP default parameters, which are just
adapted from land-based applications, may not be applicable for
underwater images. Such constraint limits the capability of this
restoration algorithm to improve the quality of an underwater image; the
values of these parameters must be searched for each underwater image. A
proposed approach on the parameter values assignment problem is to
conduct an optimized search based on Genetic Algorithm. The presentation
of this proposed approach focuses on the Genetic Algorithm processes:
chromosome encoding, fitness function development, and selection,
mutation, and crossover, to perform an effective search of the best
solution out of a pool of possible solutions. Qualitative and
quantitative evaluations show that utilization of optimized combination
of DCP parameters, achieves images of higher quality in comparison to
the utilization of established default DCP parameters.
|
|
ThB42 |
L-2 |
R2: Robotics, Control Systems & Theory |
Regular Session |
Chair: Kobayashi, Taisuke | Nara Institute of Science and Technology |
|
13:30-13:45, Paper ThB42.1 | |
Model Predictive Landing Control of an Unmanned Aerial Vehicle via Partial Feedback Linearization |
|
Zhou, Yang | Ritsumeikan University |
Takaba, Kiyotsugu | Ritsumeikan Univeristy |
Ohashi, Asuka | Ritsumeikan University |
Keywords: Robotics, Control Systems & Theory, Aerospace Technology
Abstract: This paper
introduces a model predictive control approach of an unmanned aerial
vehicle (UAV) with the aid of a feedback linearization.As is well
known,the feedback linearization is one of the effective techniques to
cope with the nonlinearity of dynamical systems.Since the UAV is an
underactuated nonlinear system, it is impossible to exactly linearize
the dynamics of the UAV.Therefore, we take an approach to linearize only
the translational motion, and then apply the linear optimal control to
it.However, the UAV is easily affected by wind disturbances in an actual
environment. A model predictive control is proposed to cope with the
disturbances. We apply this approach to a landing control of the UAV to a
moving ground vehicle.The effectiveness of the proposed method is
verified by numerical simulations.
|
|
13:45-14:00, Paper ThB42.2 | |
Passive Knee Exoskeleton Using Brake Torque to Assist Stair Ascent |
|
Chaichaowarat, Ronnapee | Chulalongkorn University |
Macha, Vidyaaranya | Indian Institute of Technology Ropar |
Wannasuphoprasit, Witaya | Chulalongkorn University |
Keywords: Robotics, Control Systems & Theory, Biomedical Engineering
Abstract: Manipulating
mechanical energy intelligently, passive exoskeletons can improve the
energy efficiency of cyclic human motions. Aiming to reduce the energy
cost of knee moment during stair ascent, this paper presents a concept
of brake-torque support activated when the knee moment is required in
the opposite direction to the angular velocity of the knee joint.
Integrating an electromagnetic brake to a crossing four-bar knee joint
with a compact design, the passive knee exoskeleton enables the
polycentric knee center of rotation covering the wide range of knee
angle during stair ascent. For preliminary validation, the surface
electromyography (EMG) of rectus femoris (RF) and biceps femoris (BF)
were studied on a healthy male volunteer wearing the exoskeleton on his
right leg. The reduction of peak muscle activity is observed as the
brake torque is applied during knee extension while the knee flexion
moment is required.
|
|
14:00-14:15, Paper ThB42.3 | |
Soil Fertilizer Recommendation System Using Fuzzy Logic |
|
Haban, Jenskie Jerlin | De La Salle University |
Puno, John Carlo | De La Salle University |
Bandala, Argel | De La Salle University |
Dadios, Elmer | De La Salle University |
Sybingco, Edwin | De La Salle University |
Billones, Robert Kerwin | De La Salle University |
Keywords: Robotics, Control Systems & Theory, Circuits and Systems
Abstract: Soil nutrients
and season have direct impact on the growth and yield of a crop.
Deficiency on the nutrient level of the soil may result to plant disease
while applying excessive amount of soil fertilizer on the other hand,
may also cause negative results to the development of the crop.
Nutrients on the soil also changes as the season changes from wet season
to dry season. This study aims to develop a fuzzy logic-based program
that will provide an appropriate amount of fertilizer to soil. The
parameters such as season, nitrogen, phosphorus and potassium level are
the inputs used on the fuzzy logic system. The researchers proposed four
kinds of fertilizer to use in this paper such as Complete, Urea,
Solophos and Muriate of Potash. Combination and amount of these
fertilizers will be based on the input parameters and fuzzy rules. These
soil fertilizer recommendations can be used for rice in an inbred light
soil.
|
|
14:15-14:30, Paper ThB42.4 | |
Human Presence Detection Using Ultra Wide Band Signal for Fire Extinguishing Robot |
|
Bandala, Argel | De La Salle University |
Sybingco, Edwin | De La Salle University |
Maningo, Jose Martin | De La Salle University |
Dadios, Elmer | De La Salle University |
Isidro, Gann Isaac | De La Salle University |
Jurilla, Rocez Deanne | De La Salle University |
Lai, Chia-Yu | De La Salle University |
Keywords: Robotics, Control Systems & Theory, Computer Architecture & Systems, Circuits and Systems
Abstract: Fire incidents
often result to associated deaths, injuries, and losses occurring
structures and properties, particularly in homes every year. In this
study, the researchers proposed a 4-wheeled fire extinguishing robot
with the ability to detect human presence in the area even when there is
fire. Multiple sensors are utilized in this study to detect nearby
flame, smoke, temperature and humidity, and obstacles through
integration with Arduino and Raspberry Pi. The proposed robot is
remotely controlled by the user over Wi-Fi through the graphical user
interface created by the researchers in Python for easy monitoring of
data and control. A camera is also mounted to the robot for surveillance
purposes. The human detection system of the robot is implemented
through using ultra-wide band radar (UWB) by utilizing the X4M300
presence sensor, which could detect human presence based on their
respiration movement. Initial testing and four experiments were
conducted to test the radar sensor's capabilities compared to the
existing methods of human detection. The researchers yielded an accuracy
of 97.29% in the testing of human detection system, proving that the
implementation of UWB radar sensor is successful.
|
|
14:30-14:45, Paper ThB42.5 | |
Improvement of Adsorption Force of an EPM-Based Adsorb Device by a Suspension |
|
Yamazaki, Kenta | Kyoto Institute of Technology |
Higashi, Yoshiyuki | Kyoto Institute of Technology |
Masuda, Arata | Kyoto Institute of Technology |
Miura, Nanako | Kyoto Institute of Technology |
Keywords: Robotics, Control Systems & Theory, Disasters and Humanitarian Technology
Abstract: This paper
shows the composition and evaluation of a suspension to ensure that an
EPM-equipped inspection Unmanned Aerial Vehicle (UAV) is adsorbed to the
bridge. Because an Electro-Permanent Magnet (EPM) adsorbs by magnetic
force, if the contact area is small, or if a gap exists between it and
magnetic material as the target, the adsorption force decreases when EPM
is magnetized and sticks. As a result, inspection work cannot be
performed safely, and there is a risk of crashing during inspection
work. Therefore, we developed a suspension that allows the EPM to be
horizontally attracted to the EPM even if there is a step on the object
to be attracted. The effect of the suspensions was confirmed by
comparing the adsorption of EPMs on steps with and without suspensions.
|
|
ThB43 |
L-3 |
ML10: Machine Learning, Cloud and Data Analytics |
Regular Session |
Chair: Tokuda, Tomoki | ATR |
|
13:30-13:45, Paper ThB43.1 | |
Development of a Vehicle Routing System for Delivery Services |
|
Tagorda, Ian Paolo | University of the Philippines - Diliman |
Calata, Lloyd Elwyn | University of the Philippines - Diliman |
Limjoco, Wilbert Jethro | University of the Philippines - Diliman |
Dizon, Carl | University of the Philippines - Diliman |
Keywords: Software & Database Systems, Machine Learning, Cloud and Data Analytics
Abstract: With the
increasing ubiquity of e-commerce also arises the need for faster and
more efficient methods to deliver goods from one place to another. Such
systems rely heavily on vehicle routing systems, which couriers usually
follow to determine their route for the day. In some countries, vehicle
routing systems are managed manually, which either depends on the
couriers themselves or on another person called a dispatcher, which is
prone to errors and may not be very efficient. Hence, this study
proposes a two-phase heuristic algorithm that will serve as the
"dispatcher" for couriers to efficiently service all the customers
within their area. A pairing of clustering and routing algorithms that
solve the travelling salesman problem the best was found. Results show
that using k-means as the clustering algorithm and a genetic algorithm
with some modifications yield the best route recommendations in terms of
total cost and processing time. It also is more efficient than the
manual method with a savings of 16.5%
|
|
13:45-14:00, Paper ThB43.2 | |
Unsupervised Clustering Based on Feature-Value / Instance Transposition Selection |
|
Kusaba, Akira | Kyushu University |
Hashimoto, Takako | Chiba University of Commerce |
Shin, Kilho | Gakushuin University |
Shepard, David | Evidation Health |
Kuboyama, Tetsuji | Gakushuin University |
Keywords: Machine Learning, Cloud and Data Analytics
Abstract: This paper
presents FITS, or Feature-value / Instance Transposition Selection, a
method for unsupervised clustering. FITS is a tractable, explicable
clustering method, which leverages the unsupervised feature value
selection algorithm known as UFVS in the literature. FITS combines
repeated rounds of UFVS with alternating steps of matrix transposition
to produce a set of homogenous clusters that describe data well.By
repeatedly swapping the role of feature and instance and applying the
same selection process to them, FITS leverages UFVS's speed and can
perform clustering in our experiments in tens milliseconds for datasets
of thousands of features and thousands of instances.We performed feature
selection-based clustering on two real-world data sets. One is aimed at
topic extraction from Twitter data, and the other is aimed at gaining
awareness of energy conservation from time-series power consumption
data. This study also proposes a novel method based on iterative feature
extraction and transposition. The effectiveness of this method is shown
in an application of Twitter data analysis. On the other hand, a more
straightforward use of feature selection is adopted in the application
of time series power consumption data analysis.
|
|
14:00-14:15, Paper ThB43.3 | |
A High-Performance Deep Learning Architecture for Host-Based Intrusion Detection System |
|
Lee, Tsern-Huei | National Chiao Tung University |
Huang, Hsiao Yen | National Chiao Tung University |
Juang, Cheng | Ming Hsin University of Science and Technology |
Keywords: Machine Learning, Cloud and Data Analytics
Abstract: Host-based
intrusion detection system (HIDS) is a necessary component for network
security, especially when more and more data are encrypted which makes
network-based intrusion detection system lose its functionality of
packet content inspection. After many years of research, it is widely
acknowledged that system calls are the preferred data source for HIDS.
In a recent paper, a novel semantic analysis approach was proposed and
shown to achieve the best performance, as compared with various previous
syntactic analysis schemes. The performance difference is profound for
modern attacks. However, the semantic analysis approach requires
considerable computational complexity. In this paper, we present a deep
learning architecture which requires no data pre-processing and is easy
to train. Experimental results show that our design has a better
performance than the semantic analysis approach.
|
|
14:15-14:30, Paper ThB43.4 | |
An
Embedded System for Image-Based Crack Detection by Using Fine-Tuning
Model of Adaptive Structural Learning of Deep Belief Network |
|
Kamada, Shin | Prefectural University of Hiroshima |
Ichimura, Takumi | Prefectural University of Hiroshima |
Keywords: Machine Learning, Cloud and Data Analytics
Abstract: Deep learning
has been a successful model which can effectively represent several
features of input space and remarkably improve image recognition
performance on the deep architectures. In our research, an adaptive
structural learning method of Restricted Boltzmann Machine (Adaptive
RBM) and Deep Belief Network (Adaptive DBN) have been developed as a
deep learning model. The models have a self-organize function which can
discover an optimal number of hidden neurons for given input data in a
RBM by neuron generation-annihilation algorithm, and can obtain an
appropriate number of RBM as hidden layers in the trained DBN. The
proposed method was applied to a concrete image benchmark data set SDNET
2018 for crack detection. The dataset contains about 56,000 crack
images for three types of concrete structures: bridge decks, walls, and
paved roads. The fine-tuning method of the Adaptive DBN can show 99.7%,
99.7%, and 99.4% classification accuracy for test dataset of three types
of structures. In this paper, our developed Adaptive DBN was embedded
to a tiny PC with GPU for real-time inference on a drone. For fast
inference, the fine tuning algorithm also removed some inactivated
hidden neurons to make a small model and then the model was able to
improve not only classification accuracy but also inference speed
simultaneously. The inference speed and running time of portable battery
charger were evaluated on three kinds of Nvidia embedded systems.
|
|
14:30-14:45, Paper ThB43.5 | |
Detection of Heart Arrhythmia Using Hybrid Neural Networks |
|
Kommireddy, Shreeven | Mahindra Ecole Centrale |
Pandey, Piyush Raj | Mahindra Ecole Centrale |
Neelisetti, Raghu Kisore | Mahindra Ecole Centrale - Computer Science Department |
Keywords: Machine Learning, Cloud and Data Analytics
Abstract: This research
presents a new approach to Heart Arrhythmia classification with the help
of neural networks. A victim diagnosed with heart arrhythmia has an
irregular beat pattern. Heart arrhythmia is harmless in most situations
however, beyond a certain limit of irregularity, it causes severe and
critical complications. Heart arrhythmia can be detected at the early
stages. Doctors can determine symptoms of the patient and prevent
further damage. Tests have been done to check the validity of neural
networks for this job, and the attempted tool used in this paper is the
same. The purpose of this research is to increase the accuracy,
precision classifying ability while conserving the generalized
pre-processing method. This research attempts to find a
correlation-based feature selection. Our proposed algorithm exceeds a
range of acceptance, in detecting an array of heart arrhythmias from an
electrocardiogram recording used with a single lead wearable monitor.
The data set used is the MIT-BIH corpus which contains ECG recordings of
unique patients. On this data set, the 5-layered neural network which
was trained maps a sequence of ECG samples to a sequence of rhythm
classes. We have created a hybrid neural network. Through the stacking
of the neural networks, we were able to achieve an accuracy of about
ninety-eight percent which was the best-case scenario.
|
|
14:45-15:00, Paper ThB43.6 | |
Indian Stock Market Prediction Using Deep Learning |
|
Maiti, Ayan | National Institute of Technology Karnataka |
Shetty, Pushparaj | National Institute of Technology Karnataka |
Keywords: Machine Learning, Cloud and Data Analytics
Abstract: In this paper,
we predict the stock prices of five companies listed on India痴 National
Stock Exchange (NSE) using two models- the Long Short Term Memory
(LSTM) model and the Generative Adversarial Network (GAN) model with
Long Short-Term Memory (LSTM) as the generator and a simple dense neural
network as the discriminant. Both models take the online published
historical stock-price data as input and produce the prediction of the
closing price for the next trading day. To emulate the thought process
of a real trader, our implementation applies the technique of rolling
segmentation for the partition of training and testing dataset to
examine the effect of different interval partitions on the prediction
performance.
|
|
ThB44 |
L-4 |
SIP6: Signal and Image Processing |
Regular Session |
Chair: Bandala, Argel | De La Salle University |
|
13:30-13:45, Paper ThB44.1 | |
A
Novel Weakly-Supervised Proposal Method Based on Vector Perceptron and
Unbalanced Action-Background Self-Supervised Boundary Inferring |
|
Huang, Xiaoping | South China Uinversity of Technology |
Ma, Lihong | South China Univ. of Tech |
Tian, Jing | National University of Singapore |
Keywords: Signal and Image Processing, Machine Learning, Cloud and Data Analytics
Abstract: In
weakly-supervised temporal action detection, only the category labels of
actions could be used for training. Without any temporal boundary
annotations in an untrimmed video, it's a challenge to propose the
starter and the end of an action (named as the proposal). In this paper,
we focus on the temporal differences of visual contents which may
change significantly when motion occurs and carry the information of
temporal boundaries. First, a motion-propagating feature vector (MPV) is
designed to describe different kinds of temporal difference in a video.
Second, we propose a temporal difference perceptron (TDP). According to
the MPV, TDP computes an attention to each video snippet to infer
action areas. However, the TDP only concentrates on the most
discriminate action areas and the attentions fail in slow actions. To
solve these, an unbalanced action-background self-supervised loss (USSL)
is custom to tune the distribution of activated areas. With an
unbalanced constraint, all forward change and some backward change
action regions are activated. And the self-supervised attention labels
force the model to widen the gap between background and action in
feature space, contributing to boundary inferring. Our method achieves
comparable performance with the state-of-arts, that outperforms STPN, an
algorithm without temporal difference message, more than 2% mAP on
THUMOS14, and is 20.8%(mAP@tIou=0.5) higher than UntrimmedNet on
ActivityNet1.2 that used temporal difference roughly.
|
|
13:45-14:00, Paper ThB44.2 | |
Impact of Pre-Processing on the Performance of Text-Independent Speaker Recognition System |
|
Palathinkal, Joshua Roy | Indian Institute of Engineering Science and Technology, Shibpur |
Vaddi, Chandra Sekhar | Defence Research and Development Organization |
Keywords: Signal and Image Processing, Machine Learning, Cloud and Data Analytics
Abstract: In this paper,
we explore the effects various pre-processing techniques have on the
performance of a text-independent speaker recognition system. The
methods used in these analysis are Spectral Subtraction Method (SSM),
Spectral Subtraction Method using Oversubtraction (SSMO), Wiener
Filtering and Minimum Mean-Square-Error Short-Time Spectral Amplitude
Estimator (MMSE-STSA), followed by a recently proposed
voice-activity-detection (VAD) algorithm. For this analysis, we extract
the Mel-frequency cepstral coefficients (MFCCs) features to distinguish
and verify the identity of various individuals using two different
modelling methods, namely, (i) Gaussian Mixture Model (GMM) [using
Expectation-Maximization (EM) algorithm] and (ii) Vector Quantization
(VQ) [using Linde-Buzo-Gray algorithm], by mapping and classifying the
MFCC features. The focus of this work is not to explore the latest or
best performance algorithm for a speaker recognition system, but rather a
robust and stable, hardware friendly, real-time system, specifically
for internet-of-things (IoT) applications. For this purpose, we have
manually collected data recorded by different individuals, using
different devices, in different environments, from various locations in
India. The results indicate that among the eight models trained using
pre-processing techniques with 47 users enrolled, six of them gave a
recognition accuracy of more than 95%, with the best accuracy of 97.8%.
|
|
14:00-14:15, Paper ThB44.3 | |
Smart Defect Detection and Sortation through Image Processing for Corn |
|
Montaムez, John Joshua | Bicol State College of Applied Sciences and Technology |
Keywords: Signal and Image Processing, Machine Learning, Cloud and Data Analytics
Abstract: This research
aims to develop smart defect detection and sortation for corn with the
use of image processing. This project helps to speed up manual
inspection to improve productivity and reduce the time consumed by
farmers in sorting. The defect detection process is done through image
processing using an open computer vision library and Python. The corn is
manually placed in the roller conveyor, passing under a camera
acquiring the real-time image. The sorting process eliminates the
damaged corn from those that were in good condition. The system acquires
real-time image data from a camera feed to a computer for analyzing
purposes. Images were scanned as the corn ear was traveling through the
conveyor. A program that effectively analyzes acquired corn images'
required features was developed in Python using the Open Computer Vision
(OCV) library. A conveyor that comes with a built-in corn sortation
mechanism was controlled by Raspberry Pi that directs the corn to its
desired group. On the evaluation of the accuracy of the system, a series
of trials were conducted. The result of the evaluation yielded a 92%
success rate in terms of defect detection and sortation. Revisions were
made after the initial testing of the project. Problems and its causes
were identified to improve the performance of the whole system.
|
|
14:15-14:30, Paper ThB44.4 | |
Ensemble of Winter's Belief Based Frameworks for Hyperspectral Endmember Extraction |
|
Shah, Dharambhai | Institute of Technology, Nirma University |
Zaveri, Tanish | Institute of Technology, Nirma University |
Trivedi, Y N | Institute of Technology, Nirma University |
Keywords: Signal and Image Processing, Machine Learning, Cloud and Data Analytics, Aerospace Technology
Abstract: The goal of
Hyper-Spectral Unmixing (HSU) is to decompose HSC imagery into a group
of pure material spectral signatures also known as endmember signatures
and fractional proportions weights that characterize the contribution of
each endmember in forming a sample. In 1999, Winter痴 proposed an idea
to address HSU that considers vertices of a simplex whose volume is
maximum as pure pixel vectors. In hyperspectral remote sensing, this
belief has much influence on HSU, especially on endmember extraction
methods. Moreover, this belief has inspired much attention, resulting in
various endmember extraction frameworks such as Simplex Growing
Algorithm (SGA), N-point FINDeR (NFINDR), Alternating Volume
Maximization (AVMAX), Successive Volume Maximization (SVMAX). In this
paper, we propose an ensemble of these frameworks intending to utilize
the best part of the result of each framework. The proposed ensemble
framework uses a majority voting approach. Our experiments, applied on
four hyperspectral datasets (Cuprite, Urban, Samson and Jasper), expose
that the ensemble framework by majority voting can provide efficient and
competitive performance compared to individual winter's belief-based
endmember extraction frameworks.
|
|
14:30-14:45, Paper ThB44.5 | |
Tomato
Septoria Leaf Spot Necrotic and Chlorotic Regions Computational
Assessment Using Artificial Bee Colony-Optimized Leaf Disease Index |
|
Concepcion II, Ronnie | De La Salle University |
Lauguico, Sandy | De La Salle University |
Dadios, Elmer | De La Salle University |
Bandala, Argel | De La Salle University |
Sybingco, Edwin | De La Salle University |
Alejandrino, Jonnel | De La Salle University |
Keywords: Signal and Image Processing, Machine Learning, Cloud and Data Analytics, Social Implications of Technology
Abstract: Visual
inspection of plant health status and disease severity may yield
subjective assessments due to error-prone sphere of colors and textures
as affected by angular photosynthetic light source and the complexity of
chlorosis. Quantification of damages on leaves due to destructive
diseases is paramount for plant and pathogen interactions. To address
this challenge, the proposed solution is the integration of computer
vision and computational intelligence for tomato Septoria leaf spot
necrotic and chlorotic region computational assessment. Dataset contains
healthy and diseased tomato leaves that were captured individually.
Non-vegetation pixels removal was done using CIELab color space. RGB
color components and five Haralick texture features were extracted from
the segmented leaf. Hybrid neighborhood component analysis and ReliefF
algorithm were employed to select the important predictors resulting to
RGB-entropy vector. A new tomato leaf disease index (tomLDI) optimized
using artificial bee colony (ABC) was developed by normalizing visible
red reflectance, and introducing red-green and red-blue reflectance
ratios to enhance Septoria leaf spots pixels and reducing sensitivity to
healthy green pixels. KNN bested classification tree, linear
discriminant analysis and Na阮e Bayes in detecting Septoria leaf disease
with accuracy of 97.46%. Deep transfer image regression was tested
using raw infected leaf images and the tomLDI transformed colored
channels through MobileNetV2, ResNe
|
|
14:45-15:00, Paper ThB44.6 | |
Age Progression Using Generative Adversarial Networks |
|
Madhukar, Pallavi | PES University |
Chetan, Rachana | PES University |
Prasad, Supriya | PES University |
Shayan, Mohamed | PES University |
Krupa, Niranjana | PES University |
Keywords: Signal and Image Processing, Machine Learning, Cloud and Data Analytics, Social Implications of Technology
Abstract: This study
presents a technique to generate face age progression by adopting a
conditional generative adversarial network based approach. The best
model resulting from a five-fold cross validation has an accuracy of
91.93%, False Omission Rate of 0.45% and Negative Prediction Value of
99.55%. Building on prior work, this paper has three contributions.
First, the use of uneven age clusters is presented to account for more
rapid and drastic ageing in babies and toddlers than older individuals.
Second, perceptual losses rather than per-pixel losses are considered to
enable identity preservation. Third, a facial recognition system is
applied to verify the identity of individuals upon ageing. Identity
preservation was achieved and confirmed, with a facial recognition
accuracy of 92.4%. Visual fidelity was also confirmed, with 95.2%
subjects identifying aging in the conducted survey.
|
|
ThB45 |
L-5 |
ME2: Multimedia Engineering |
Regular Session |
Chair: Chen, Na | Nara Institute of Science and Technology |
|
13:30-13:45, Paper ThB45.1 | |
Game-Based Learning Tool for Photography |
|
Romlamduan, Napat | Mahidol University |
Kusakunniran, Worapan | Mahidol University |
Keywords: Multimedia Engineering, Engineering Education
Abstract: Nowadays, it is
convenient to gain new knowledge and skills by learning from the
internet. However, some contents may be hard to understand just by
reading texts. Photography knowledge is one of those, in which learner
may need a large amount of time and cost to practice it. Therefore, this
paper provides an alternative way of a game-based learning tool. It
simulates a camera into a game combining with a story and some
challenges of engagement. It is designed such that a photography
knowledge can be delivered to learners through this developed game. To
validate this assumption, 25 participants are asked to join our
experiment. They are asked to do a pre-test, play our game, and do a
post-test. Then, they are asked to answer questionnaires regarding
effectiveness and benefits of the proposed game. It is shown that the
participants could improve their scores from the pre-test of 44% to the
post-test of 89%, regarding the understanding of photography knowledge.
Also, the questionnairesresults show that our game could help the
participants to gain the knowledge.
|
|
13:45-14:00, Paper ThB45.2 | |
Human Motion Recognition by Three-View Kinect Sensors in Virtual Basketball Training |
|
Baoqi, Yao | Beijing University of Posts and Telecommunications |
Gao, Hui | Beijing University of Posts and Telecommunications |
Su, Xin | Tsinghua University |
Keywords: Multimedia Engineering, Machine Learning, Cloud and Data Analytics
Abstract: In recent
years, human action recognition has received a considerable amount of
research attention because of it's potential in a variety of
applications, such as video surveillance, human-computer interaction,
and virtual reality (VR). However, many researches on human action
recognition performed in single-camera or double-camera system, which
achieve reduced performance due to vulnerability to partial occlusion
and missrecognition of back. Some works on human action recognition use
multiple cameras but are too complex for practical application. In this
paper, we propose a new human action recognition system using triple
Kinect sensors for VR application. Particularly, we design a mark
detection method to determine the front of user and fusion skeleton data
in real time. Features are extracted from three-dimensional (3D)
skeleton data sequences, and divided into five parts according to body
parts. A classification model based on the part-aware long short-term
memory networks is proposed to recognize human motion. Finally, we
demonstrate the system with a virtual reality basketball application and
the results of experiment validate the feasibility of the proposed
system.
|
|
14:00-14:15, Paper ThB45.3 | |
Influences of Network Delay on Cooperative Work in Networked Virtual Environment with Haptics |
|
Aung, Su Thandar | University of Computer Studies, Yangon |
Ishibashi, Yutaka | Nagoya Institute of Technology |
Mya, Khin Than | University of Computer Studies, Yangon |
Watanabe, Hitoshi | Tokyo University of Science |
Huang, Pingguo | Gifu Shotoku Gakuen University |
Keywords: Multimedia Engineering, Robotics, Control Systems & Theory
Abstract: In this paper,
we investigate influences of the network delay on haptic cooperative
work in a networked virtual environment by experiment. In the work, two
users carry an object cooperatively in a networked maze system by
manipulating haptic interface devices. We handle two carrying methods in
our experiment. In one method, the two users hold the object. In the
other method, one user holds the object and the other user pushed it. We
measure the operation time and reaction force, and we compare them
between the two methods. Experimental results demonstrate that the
former method has smaller average operation time than the latter method,
but the average reaction force is larger in the former. Keywords湧etworked virtual environment, Haptics, Cooperative work, Maze system, Network delay, Experiment
|
|
14:15-14:30, Paper ThB45.4 | |
Hardware Parallel Processing of 3 ラ 3-Pixel Image Kernels |
|
Kho, Daniel, C.K. | Multimedia University |
Ahmad Fauzi, Mohammad Faizal | Multimedia University |
Lim, Sin Liang | Multimedia University |
Keywords: Multimedia Engineering, Signal and Image Processing, Circuits and Systems
Abstract: Video
processing usually requires one to read in an entire image into a
framebuffer, usually taking the form of random access memory (RAM). For
kernel-based image and video processing, square-sized kernels are then
extracted from this framebuffer, typically in 3ラ3-pixel sizes, though
other sizes are also common. To meet the demand for ever higher image
resolutions, larger and larger framebuffer RAM memories are required. While
it is not feasible for software to read and process parts of an image
quickly and efficiently enough due to the high speed of incoming video, a
hardware-based video processing solution poses no such limitations.
RAM-based framebuffers can also be found on hardware-based video
processing systems, however such designs are not leveraging the full
power and potential of processing image kernels with digital hardware.
This paper introduces hardware techniques to read and process kernels
without the need to store the entire image frame. This reduces the
memory requirements significantly without loss of quality to the
processed images.
|
|
14:30-14:45, Paper ThB45.5 | |
Illegal Logging Listeners Using IoT Networks |
|
Srisuphab, Ananta | Mahidol University |
Kaakkurivaara, Nopparat | Kasetsart University |
Silapachote, Piyanuch | Mahidol University |
Tangkit, Kitipong | Kasetsart University |
Meunpong, Ponthep | Kasetsart University |
Sunetnanta, Thanwadee | Mahidol University |
Keywords: Devices, Materials & Processing, Signal and Image Processing
Abstract: Protecting and
increasing worldwide green space have been an international effort.
Individuals and organizations are encouraged to plant urban trees and to
get involved in many reforestation and restoration projects. Offsetting
these much needed plans to save the forests is illegal logging. Trees
that have grown for many years, some are protected resources inside
restricted areas, are felled and the wood is smuggled. Watching for
these illegal activities is very difficult and also very dangerous. It
is quite impossible for rangers to patrol every entry and exit point of
forests that cover thousands of squared kilometers. Applying Internet of
Things technology to ecological forestry, we are proposing integrating
sound acquisition networks and acoustic signal analyzers to enhance the
robustness of an already successful camera-based surveillance solution
that is also equipped with a global positioning system tracker. Our
listener devices record sounds of the forest and periodically send it to
a cloud storage over cellular networks. The device is affordable, the
system is small and portable, and the network is flexibly extensible.
From the data, acoustic features are extracted and visualized. The
Mel-frequency cepstral coefficients of the signals have exhibited
promising distinctiveness for detection of illegal chainsaw activities
in the wild.
|
|
14:45-15:00, Paper ThB45.6 | |
Water Level Detection from CCTV Cameras Using a Deep Learning Approach |
|
Borwarnginn, Punyanuch | Mahidol University |
Haga, Jason | National Institute of Advanced Industrial Science and Technology |
Kusakunniran, Worapan | Mahidol University |
Keywords: Disasters and Humanitarian Technology, Signal and Image Processing, Machine Learning, Cloud and Data Analytics
Abstract: Natural
disasters are a global problem that causes widespread losses and damage.
A system to provide timely information is required in order to help
reduce losses. Flooding is one of the major natural disasters that
requires a monitoring and detection system. The traditional flood
detection systems use remote sensors such as river water levels and
rainfall to provide information to both disaster management
professionals and the general public. There is an attempt to use visual
information such as CCTV cameras to detect extreme flooding events;
however, it requires human experts and consistent attention to monitor
any changes. In this paper, we introduce an approach to the automatic
river water level detection using deep learning to determine the water
level from surveillance cameras. The model achieves 93% accuracy using a
single camera location and 83% accuracy using multiple camera
locations.
|
|
ThB46 |
L-6 |
P6: Power & Energy |
Regular Session |
Chair: Vo, Quoc Trinh | Nara Institute of Advanced Science and Technology |
|
13:30-13:45, Paper ThB46.1 | |
Design Consideration of Self-Oscillating Voltage Booster Including JFET Characteristics for Energy Harvesting Applications |
|
Onishi, Kosuke | Kyushu Institute of Technology |
Matsuo, Masahiro | Ricoh Electronic Devices Co., Ltd |
Abe, Seiya | Kyushu Institute of Technology |
Keywords: Power & Energy, Circuits and Systems
Abstract: An Energy
Harvesting (E.H.) has been paid attention for the field of Internet of
Things (IoT) in recent years. The wireless sensor as a key component of
IoT can be realized battery less operation by using E.H technology. The
oscillating circuit with a step-up transformer is used as a voltage
booster for E.H. The voltage booster has not been fully analyzed the
influences of the circuit parameters yet. In this paper, the influence
of JFET parameters such as parasitic capacitance and threshold voltage
are analyzed, and the self-oscillating voltage booster is designed
including the JFET parameters. As a result, it is clarified that the
large additional capacitance connected into gate-source terminal of JFET
can be remarkably reduced the influence of input capacitance of JFET.
Moreover, the relationship between threshold voltage and amplitude
condition are clarified. Finally, the validity of the analysis and
design method are experimentally confirmed.
|
|
13:45-14:00, Paper ThB46.2 | |
Power Flow Reversal Control of Hybrid HVDC Transmission System with LCC and FBMMC |
|
Shi, Yuqian | Shanghai Jiao Tong University |
Zhu, Miao | Shanghai Jiao Tong University |
Chen, Yang | Shanghai Jiao Tong University |
Cai, Xu | Shanghai Jiao Tong University |
Keywords: Power & Energy, Circuits and Systems
Abstract: A power flow
reversal control strategy is proposed for bidirectional power
transmission of LCC and FBMMC hybrid HVDC transmission system. The power
flow reversal sequence is designed with integrating DC voltage control
and active current control firstly. The proposed dual-loop control
strategy for FBMMC effectively eases the FBMMC AC-side power and
submodule capacitor voltage fluctuations during the power flow reversal
process. The inner current control loop of FBMMC model based on the
sliding mode variable structure control theory is established to
suppress the influence of the reference value mutation on the system.
Finally, a simulation model of hybrid HVDC transmission system with
12-pulse LCC and 51-level FBMMC is built in MATLAB/Simulink, and the
improvement of the system dynamic performance is demonstrated.
|
|
14:00-14:15, Paper ThB46.3 | |
An Improved Triple Interline DC Power Flow Controller for Bidirectional Power Control |
|
Yi, Jing | Shanghai Jiao Tong University |
Zhu, Miao | Shanghai Jiao Tong University |
Zhong, Xu | Shanghai Jiao Tong University |
Wang, Han | Shanghai Jiao Tong University |
Cai, Xu | Shanghai Jiao Tong University |
Keywords: Power & Energy, Circuits and Systems
Abstract: For the
operational safety and stability of electrical system, effective power
flow control in meshed high voltage direct current (HVDC) grids makes
sense. In order to realize the objective of controlling several lines
simultaneously and independently in a convenient way, interline dc power
flow controller (IDCPFC) has been introduced. To cover some shortages
of present IDCPFC topologies, an improved triple interline dc power flow
controller (TI-DCPFC), capable of controlling power flow of two lines
actively, has been proposed in this paper. To explain it thoroughly and
explicitly, its operational principle is illustrated combined with the
mathematic equations firstly. Then, to validate its capacity of dealing
with occasions of bidirectional flows and the dual-freedom control
function, the theoretical control strategy and simulation results in a
typical working mode are referred to comprehensively
|
|
14:15-14:30, Paper ThB46.4 | |
Development
of a High Efficiency DC-DC Converter Using Hysteretic Control for
Hydroelectric Energy Harvester in a Wireless Sensor Network |
|
Canacan, Gennylyn | FAITH Colleges |
Llanto, John Thimotee | FAITH Colleges |
Moredo, Eala Eireen | FAITH Colleges |
Santos, Adonis | FAITH Colleges |
Malabanan, Francis | FAITH Colleges |
Tabing, Jay Nickson | FAITH Colleges |
Geva, Sherryl | FAITH Colleges |
Keywords: Power & Energy, Circuits and Systems
Abstract: Efficiency is a
requirement when it comes to utilizing a wireless sensor network (WSN)
where hydrokinetic energy harvesting through turbine is involved. Thus,
not only WSN needs an efficient supply but also the sensors and the
storage unit which are powered up by an energy harvesting module, a
turbine DC generator. Turbine DC generator produces a low voltage and
low voltage means low power. To produce high efficiency output despite
the low power it produces, a DC-DC converter is one of the
preliminaries. DC-DC converter regulates its input coming from the
turbine DC generator and produces a more stable power supply. However,
blocks that the DC-DC converter supplies have different voltage
requirement. Therefore, the researchers will develop two DC-DC converter
topology which are the Buck Converter and the Boost Converter. On the
contrary, turbine DC generator produces varying DC supply to the boost
converter inducing noises and reducing the efficiency needed. Therefore,
to achieve a highly efficient output the device needs to be low noise.
To prevent noise from affecting the efficiency of the device, the
researchers will use a technique called Hysteretic Control (HC) of DC-DC
converter. This research intended to design a high efficiency Direct
Current-to-Direct Current Converter for hydroelectric energy harvester
in wireless sensor network. The researchers develop a DC-DC converter
with an efficiency of 80% - 95% by reducing the noise by using a
switching DC-DC converter.
|
|
14:30-14:45, Paper ThB46.5 | |
Accuracy-Configurable Low-Power Approximate Floating-Point Multiplier Based on Mantissa Bit Segmentation |
|
Li, Jie | Waseda University |
Guo, Yi | Waseda University |
Kimura, Shinji | Waseda University |
Keywords: Power & Energy, Circuits and Systems, Computer Architecture & Systems
Abstract:
Abstract湧owadays, in energy-efficient design of digital systems,
approximate computing (AC) has an increasingly important role. Due to
human perceptual limitations, redundancy in input data and so on, there
is a huge amount of applications that can tolerate errors. In this
paper, an accuracy-configurable approximate floating-point (FP)
multiplier is proposed to improve hardware consumption for such
applications. Mantissa is divided into a short exactly processed part
and a remaining approximately processed part. A new addition and
shifting method is applied to the approximate part to replace
multiplication to improve hardware performance. Experimental results
show the 4-bit exact part configuration of the proposed work ensures the
accuracy of 99.17% (MRED is 0.83%) with the reduction 67.65% of area,
16.64% of delay and 75.62% of power. The proposed work also shows good
performance in image processing and neural networks.
|
|
14:45-15:00, Paper ThB46.6 | |
Development of Energy Management Unit for Hydropower Storage of Harvesting System in Wireless Sensor Networks |
|
Dimaano, Nicolle | FAITH Colleges |
Maligalig, Xavier | FAITH Colleges |
Rosales, Marielle Angelica | FAITH Colleges |
Santos, Adonis | FAITH Colleges |
Malabanan, Francis | FAITH Colleges |
Tabing, Jay Nickson | FAITH Colleges |
Geva, Sherryl | FAITH Colleges |
Keywords: Power & Energy, Circuits and Systems, Signal and Image Processing
Abstract: One of the
basic sources of life in earth is water and as the population of humans
living in it increases, water consumption and contamination also
increases. Thus, the researchers aim to develop a sensor platform that
monitors the quality and contamination of water bodies. There are
commercially available sensor platforms that has sensor nodes which
transmit data it gathered wirelessly to a base station and are
battery-driven. However, due to lack of recharging capability of
batteries, the life of sensor platforms is shortened, and the
replacement of these batteries became a burden. Hence, this research
aims to develop and implement an energy management unit of a harvesting
system in a sensor platform through a 90nm process technology in order
to eliminate the burden of replacing the batteries and extend the sensor
platform痴 life by efficiently managing the energy sources given by the
environment and the battery.
|
|
ThC41 |
L-1 |
ENG1: Engineering Education & Engineering Management |
Regular Session |
Chair: Jinrong, Liu | University of Hong Kong |
|
15:15-15:30, Paper ThC41.1 | |
Comparative Evaluation of Assessment Activities in an Introductory Occupational Safety and Health Course |
|
Daculan, Erwin | University of San Carlos |
Keywords: Engineering Education
Abstract: This paper
emerged during the process of continuous quality improvement of an
introductory course on occupational safety and health. Did the students
correctly associate an assessment activity to the right course outcome?
The paper compared the evaluation of the students and the facilitator on
the different activities used to assess the demonstration of
abilities/outcomes attributed with the course. There were two activities
per quarter of the semester. Each activity had been scheduled to
demonstrate specific outcome indicators by the facilitator and is
described in detail. An evaluation questionnaire was provided at the end
of the semester to ascertain whether the students assessed the same
outcome indicator as what course the set for each activity. The
contention was to harmonize what outcomes the students believed they are
exhibiting and what outcomes the course had intended for such activity.
The gathered data from the evaluation questionnaire were tabulated and
then compared with the original intended outcomes by the facilitator.
Interpretation of the data was also forwarded and the path for
continuous quality improvement was drawn.
|
|
15:30-15:45, Paper ThC41.2 | |
Application of the Reverse Engineering As an Early Engineering Education |
|
Komatsu, Masaaki | National Institute of Technology |
Aburatani, Hideaki | National Institute of Technology |
Teawhim, Sanit | KOSEN-KMITL |
Keywords: Engineering Education
Abstract: The
KOSEN-KMITL, the first Japanese style KOSEN in Thailand, was established
on May 2019 in King Mongkut痴 Institute of Technology Ladkrabang
(KMITL). KOSEN-KMITL is an Engineering, Technology and Innovation
workforce development project, to support, raise the investment and
increase the industrial capacity of the Thai industry region. The
establishment of the KOSEN-KMITL , which adopts the high-quality
teaching and learning program courses under the guidance of the Japan
KOSEN system, will build and develop the skilled human resources in
engineering who can apply knowledge and skill, not only to create and
develop innovations that help solving problems, but also to make 砺alue
addedin the industrial and social sector in Thailand. The KOSEN-KMITL
course is different from the typical Engineering program in the
Universities, with its 5-year course focusing on the practical
engineering along with the theory and building knowledge based in
several areas such as social sciences and economics that will students
to understand human, public mind, social responsibility and the ability
to communicate in both English and Japanese. In the first year in
KOSEN-KMITL, students will learn 的ntroduction to Engineering Approach
的ntroduction to Engineering Design and 鏑ab workas an early
Engineering introductory education stage. These are the main subjects to
learn the concept and methodology of Engineering Approach, Engineering
Design, and various measurement technique and theoretical
|
|
15:45-16:00, Paper ThC41.3 | |
Performance Comparison between Different Rotor Configurations of PMSM for EV Application |
|
Murali, Namitha | Department of Electrical Engineering, CET, KTU |
S, Ushakumari | Centre of Excellence in Electric Vehicle, TrEST Research Park |
V.P, Mini | Department of Electrical Engineering, CET, KTU |
Keywords: Engineering Education, Engineering Management, Devices, Materials & Processing
Abstract: The modern
world is concerned about the coal reserves, hiking prices of petroleum
products, air pollution, and global warming. These all create a great
demand for Electric Vehicles (EV). The government authorities have
already initiated several projects which targeted on sales of millions
of EVs by 2025. Electric three wheelers (E-rickshaw) have a high demand
in a country like India since it痴 a common mode of transportation.
Different types of AC and DC motors are used in EV, among them Permanent
Magnet Synchronous Motor (PMSM) posses high efficiency due to its high
power density, small size, great torque inertia ratio, better overload
capacity, high output torque, and fewer losses. The torque, efficiency
and power density of PMSM is dependent upon its rotor configuration. The
paper discusses detailed design of PMSM with different rotor
configurations, especially concentrating on the design and calculation
of air gap length and permanent magnet dimensions. The designed values
are further used for the virtual development of PMSM on ANSYS MAXWELL
platform, considering seven different rotor topologies like surface-
mounted, de-centred PM, spoke type, interior PM, U-shaped, V-shaped,
modified V-shaped. A detailed comparative study based on performance
characteristics like weight, cogging torque, power, efficiency, air gap
flux density, magnetic flux density, etc. obtained from the Maxwell
platform was done. The paper concludes by the selection of modified
Vshaped rotor topology.
|
|
16:00-16:15, Paper ThC41.4 | |
Audit Pattern Optimization in Service Industry Using Six Sigma Methodology |
|
Alejandrino, Jonnel | De La Salle University |
Magno, Darlyn Jasmin | Megaworld Corporation Executive |
Concepcion, Ronnie II | De La Salle University |
Lauguico, Sandy | De La Salle University |
Tobias, Rogelio Ruzcko | De La Salle University |
Almero, Vincent Jan | De La Salle University |
Dadios, Elmer | De La Salle University |
Flores, Ramon | Laguna State Polytechnic University |
Aranas, Cheerobie | Laguna State Polytechnic University |
Keywords: Engineering Management
Abstract: Six Sigma is an
engineering systematic methodology modeled within the DMAIC course
(Define, Measure, Analyze, Improve, and Control). It is used for
improving high-risk areas that can affect the profitability of the
organization. This paper aims to prove the flexibility of Six Sigma by
increasing the efficiency rate of the internal audit team in an existing
real-estate company by targeting standard man-days or further reducing
it to less than the set timeline. The efficiency rate of the audit team
is 41.67% with average man-days of 26.9. Timeliness of audit report has
been an integral contributor of on-time resolution of non-conformities
and appropriate mitigation of high risks incurred on those findings.
Through cause-effect analysis guided by the Affinity method, probable
causes were listed down, and for the perceived causes with no measurable
data further investigation and walkthrough was done using GEMBA walk.
Data were collected and analyzed to validate these causes. Hypothesis
tests and control charts like ANOVA test, Ishikawa Diagram, and Boxplot
were used through the use of Engineering simulation programs, to analyze
the gathered data and results revealed the root causes of the problem
such as manual extraction of audit evidence, difficulty searching and
retrieving records, a variation on auditorsperformance, and
unawareness of risk management procedures among auditees. Upon
implementation of improvement solutions, the average man-days reduced
from 26.9 days to 23.8
|
|
16:15-16:30, Paper ThC41.5 | |
Conformity of Urban Development to the Approved Comprehensive Land Use Plan of Butuan City |
|
Cloma, Clint | Caraga State University |
Asube, Lorie Cris | Caraga State University |
Keywords: Engineering Management
Abstract: Abstract
Butuan City is a highly urbanized city and the regional center of the
Caraga Region. However, improper land use is a significant problem that
leads to the promulgation and implementation of the Comprehensive Land
Use Plan in 2002 to ensure sustainable urban development and proper
utilization of land resources. This study aimed to map the urban
built-up of Butuan City for the years 2003, 2011, and 2019 using
conventional machine learning classification techniques, and to quantify
its conformity to the CLUP. The researcher used Landsat Imagery to
generate the land cover of Butuan City. In this study, the Support
Vector Machine (SVM) Classification overall accuracy is 95.62%, 95.64%,
and 93.60% for 2003, 2011, and 2019, respectively. The urban built-up
conformity is 88.68%, 83.42% and 78.08% for 2003, 2011 and 2019,
respectively. This study concludes that the remote sensing technique can
help monitor urban development, and the results can help establish
strategies to promote sustainable development.
|
|
16:30-16:45, Paper ThC41.6 | |
Extraction of Starch from Cassava Tuber As a Substitute of Polyethylene in Packaging Bag |
|
Baldovino, Christine Lorraine | Pamantasan Ng Cabuyao |
Daluyin, Efraim | Pamantasan Ng Cabuyao |
Develles, Jessa | Pamantasan Ng Cabuyao |
Duyogan, Paula Jean | Pamantasan Ng Cabuyao |
Monsanto Jr., Marincho | Pamantasan Ng Cabuyao |
Sigue, Anna-liza | Pamantasan Ng Cabuyao |
Alcaraz, Jesse Ian Llyod | Pamantasan Ng Cabuyao |
Alcantara, Ramonchito | Pamantasan Ng Cabuyao |
Vanguardia, Sarah | Pamantasan Ng Cabuyao |
Keywords: Engineering Management, Biomedical Engineering
Abstract: Plastic
revolutionized the industry during its invention. It helps us not only
to have strong and flexible tools but also to have cheaper materials for
almost everything we make. Despite of its usefulness, it also has
diverse effects in our ecosystem. Hazardous chemicals leaking from
landfilled plastics runs through rivers and lakes which are primary
sources of water, aquatic animal suspect plastics as foods which make
them feel full all the time without nutrients coming from their body and
causes them to die in hunger and the birth of 溺icroplasticpollution
in the environment. Those are some of negative effects of plastics in
our environment which is critical for our future generations. The
researchers introduces a biodegradable packaging bag that focuses on
creating a product to be used as alternative to plastics and solve the
negative effects of plastics that affect environment through the use of
safe and bio-degradable materials such as natural starch from cassava,
glycerol, and many more. By mixing the right proportions of the main
materials and perfect amount of heat, the researchers achieved the best
features of the bio-degradable polymer that will be discussed later in
this paper.
|
|
ThC42 |
L-2 |
R3: Robotics, Control Systems & Theory |
Regular Session |
Chair: Jia, Haohui | Nara Institute of Science and Technology |
|
15:15-15:30, Paper ThC42.1 | |
Fuzzy Irrigation System with Rain Detection and Fertilizer Control |
|
Pareja, Michael | De La Salle University |
Bandala, Argel | De La Salle University |
Keywords: Robotics, Control Systems & Theory, Machine Learning, Cloud and Data Analytics
Abstract: Irrigation is
essential for growing crops and leads to gradual growth in the economy.
This research proposal aims to resolve the issue of scarcity and proper
water management in the tank system through the Fuzzy Irrigation System.
Fuzzy logic improves the irrigation system that includes three input
parameters, such as soil moisture, soil temperature, and the water
level. The combinations of these parameters will produce the time
duration to have an efficient flow of water to the crop fields.
Likewise, the Rain Detection Model (RDM) and the Fertilizer Control
Model (FCM) are other features that support, strengthen, and innovate
the system. The pilot test conducted by the researcher through MATLAB
simulations were performed to check the effectiveness of the proposed
system before its actual implementation.
|
|
15:30-15:45, Paper ThC42.2 | |
Adaptive Compensator of Magnetic Levitation System Using Symbolic Regression |
|
Palconit, Maria Gemel | De La Salle University |
Fuentes, Rizaldo | Teradyne |
Narvios, Wilen Melsedec | Cebu Technological University - Main Campus |
Rosales, Marife | De La Salle University |
Bandala, Argel | De La Salle University |
Dadios, Elmer | De La Salle University |
Keywords: Robotics, Control Systems & Theory, Machine Learning, Cloud and Data Analytics, Circuits and Systems
Abstract: The tuning
process for a magnetic levitation to control the object's gap from the
electromagnet is laborious and demands ample effort to obtain an
adaptive PID compensator. Hence, this study has schemed an unexplored
adaptive feedforward compensator for a 1-DOF maglev system using
equation search based on a symbolic regression through an evolutionary
algorithm. Results have shown an exceptional accuracy with an r2 of
0.9997, almost zero root mean square error (RMSE) and mean absolute
error (MAE). The approach has paved the way for an adaptive nonlinear
system requiring a highly accurate model with a baseline dataset
containing few modifiable parameters.
|
|
15:45-16:00, Paper ThC42.3 | |
AI Based Greenhouse Farming Support System with Robotic Monitoring |
|
Fernando, Sandunika | Sri Lanka Institute of Information Technology |
Kalutota Kuruwe Arachchige, Ranusha Nethmi | Sri Lanka Institute of Information Technology |
Silva, Ashen | Sri Lanka Institute of Information Technology |
Perera, Ayesh | Sri Lanka Institute of Information Technology |
De Silva, Rajitha | Sri Lanka Institute of Information Technology |
Abeygunawardhana, Pradeep | SLIIT |
Keywords: Robotics, Control Systems & Theory, Machine Learning, Cloud and Data Analytics, Signal and Image Processing
Abstract: Greenhouses
plays a major role in today痴 agriculture since farmers can grow plants
under controlled climatic conditions and can optimize production. The
greenhouses are usually built in areas where the climatic conditions for
the growth of plants are not optimal so requires some artificial setups
to bring about productivity. Automating process of a greenhouse
requires monitoring and controlling of the climatic parameters.This
paper is an attempt to minimize the cost of maintaining greenhouse
environments using new technologies. The end goal of this research an
automated system to optimally monitor and control the environmental
factors inside greenhouse by monitoring temperature, soil moisture,
humidity and pH through a cloud connected mobile robot which can detect
unhealthy plants using image processing and machine learning. The mobile
robot navigates through a predefined map of greenhouse. Database server
has created to store gathered real-time data. And the necessary
accurate data represent by using proper application for analyzing.
|
|
16:00-16:15, Paper ThC42.4 | |
Automatic Anchor Calibration for UWB-Based Indoor Positioning Systems |
|
Yasukawa, Yukiya | Kyoto Institute of Technology |
Higashi, Yoshiyuki | Kyoto Institute of Technology |
Masuda, Arata | Kyoto Institute of Technology |
Miura, Nanako | Kyoto Institute of Technology |
Keywords: Robotics, Control Systems & Theory, Wireless Communications & Networks
Abstract: In order to
navigate Unmanned Aerial Vehicles (UAVs) in GNSS-denied environments
localized indoor positioning systems is essential. One of the most
common methods for such positioning systems uses high-bandwidth radio
frequency called Ultra-wideband (UWB). The UWB radio has several
characteristics suited for indoor communication such as low
susceptibility to multipath and high immunity against wireless network
interference. Although UWB-based positioning is widely used, most of the
existing solutions require accurate coordinates of UWB nodes to be
known in advance. This put limitation on deployment of positioning
systems. In this study, we designed a method to automatically calibrate
UWB nodes positions in order to make deployment of indoor positioning
systems easier. The proposed method for auto-calibration are tested in
real indoor environments to validate effectiveness of the strategy. The
effect of node arrangement on accuracy of position estimation was
examined as well. As the result, we could improve performance of
auto-calibration by changing geometry of UWB nodes.
|
|
ThC43 |
L-3 |
ML12: Machine Learning, Cloud and Data Analytics |
Regular Session |
Chair: Tokuda, Tomoki | ATR |
|
15:15-15:30, Paper ThC43.1 | |
Ensuring Green Computing in Reconfigurable Hardware Based Cloud Platforms from Hardware Trojan Attacks |
|
Guha, Krishnendu | University of Calcutta |
Majumder, Atanu | University of Calcutta |
Saha, Debasri | University of Calcutta |
Chakrabarti, Amlan | University of Calcutta |
Keywords: Power & Energy, Machine Learning, Cloud and Data Analytics
Abstract: Deployment of
reconfigurable hardware or field programmable gate arrays (FPGAs) in
cloud platforms is the modern trend. Practical scenarios include
Amazon's EC2 F1 cloud services, Microsoft's Project Catapult and many
others. Efficient task scheduling algorithms exist that can ensure green
computing, i.e. order the operation of user tasks in the available
FPGAs in such a manner that the power dissipated is optimum. But recent
literature has exhibited eradication of the hardware root of trust,
which is not taken into account by the existing task scheduling
algorithms that can facilitate green computing. In this work, we analyze
how vulnerability in hardware like hardware trojan horses (HTH) can
increment power dissipation suddenly at runtime, without affecting the
basic security primitives like integrity, confidentiality or
availability of the system. Thus, are difficult to detect but may hamper
the system due to unnecessary high power dissipation. We also develop a
suitable runtime task scheduling algorithm which schedules the tasks at
runtime based on the dynamic status of the resources, such that the
power dissipation incurred at runtime is optimum. Finally, we also
propose a mechanism via which we can detect affected cloud resources
based on the runtime operations. We validate our proposed methodology
via simulation based experiments.
|
|
15:30-15:45, Paper ThC43.2 | |
Automatic
Pothole Classification and Segmentation Using Android Smartphone
Sensors and Camera Images with Machine Learning Techniques |
|
Thiruppathiraj, Sowbarnika | IIIT Bangalore |
Kumar, Uttam | IIIT Bangalore |
Buchke, Swapnil | IIIT Bangalore |
Keywords: Machine Learning, Cloud and Data Analytics, Signal and Image Processing
Abstract: In most
developing countries, a lag in maintaining the existing road
infrastructure often leads to development of potholes, cracks, etc. Such
irregularities in maintenance are one of the main causes of road
accidents as well as wear and tear of vehicles. This necessitates
development of automated techniques for detecting and segmenting the
potholes on roads. In this paper, a solution is proposed to detect and
segment dry and wet potholes (potholes filled with water) using
smartphone sensors and camera images. For the first set of experiments
using smartphone sensor records, various machine learning techniques
(RF, XGBoost and ANN) with balanced and imbalanced classifiers were used
to detect the potholes. A new pothole detection approach with the
sensor data is proposed using GMM (Gaussian mixture model) for a
two-class (pothole and normal) hypothesis testing problem that rendered
70% accuracy with precision and recall values ranging from 50% to 60%.
For the second set of experiments on camera images and videos captured
from a moving vehicle, popular semantic segmentation techniques like
Mask-RCNN and U-Net algorithms were applied. The results showed that
Mask RCNN rendered an overall accuracy of 80% with precision, recall and
F1 score of 0.85. The proposed framework is not only a cost effective
mechanism for detecting potholes on existing roads, its usage can also
be extended for real time pothole detection in autonomous vehicles.
|
|
15:45-16:00, Paper ThC43.3 | |
Student痴 Executive Function Assessment Tool Using Convolutional Neural Network |
|
Sanchez, Maria Trinidad Ursula | Mapua |
Singson, Lowell | Mapua |
Villaverde, Jocelyn | Mapua University |
Keywords: Machine Learning, Cloud and Data Analytics, Signal and Image Processing
Abstract: Executive
function and Academic stress is a psychological factor that affects the
mental process of human beings such as their cognitive-behavioral
control: the identification and effective monitoring of activities that
promote the achievement of chosen objectives. Left unattended, it
impairs someone痴 ability to control their behavior and negatively
affects their mental development. The developed system makes use of a
wireless body sensor network that obtains ECG signals from the student
in real-time. Which the system condition are analysed or evaluated. The
testing results which are demonstrated tested that from the CNN and ECG
Analysis, 61.11% accuracy of emotion recognition was still achieved.
Games were used to determine the WM, Flexibility, and Self-Control
levels and were similar to the neuropsychological test results.
|
|
ThC45 |
L-5 |
ME3: Multimedia Engineering |
Regular Session |
Chair: Chen, Na | Nara Institute of Science and Technology |
|
15:15-15:30, Paper ThC45.1 | |
Brightness Preserving Generalized Histogram Equalization |
|
Tanaka, Hideaki | Tokyo City University |
Taguchi, Akira | Tokyo City University |
Keywords: Signal and Image Processing, Multimedia Engineering
Abstract: Histogram
equalization (HE) is the one of the simplest and most effective methods
for contrast enhancement. It can automatically define the gray-level
transformation function based on the distribution of gray-level included
in the image. However, HE fails to produce satisfactory results for
broad range of low-contrast images because the HE does not use a spatial
feature included in the input image. The differential gray-level
histogram (DH) which is contained edge information of the input image,
were defined and the differential gray-level histogram equalization
(DHE) has been proposed. The DHE shows better enhancement results
compared to the HE result for many kinds of images. In this paper, we
propose a generalized histogram equalization (GHE) method including the
HE and the DHE. In GHE, the histograms is created using power of
gradient including a special feature. In the HE, the mean brightness of
the enhancement image cannot be controlled. On the other hand, the GHE
can control the mean brightness of the enhancement image by changing the
power, thus, the mean brightness of the input image is perfectly
preserved.
|
|
15:30-15:45, Paper ThC45.2 | |
HemoSmart: A Non-Invasive, Machine Learning Based Device and Mobile App for Anemia Detection |
|
Jayakody, Anuradha | Sri Lanka Institute of Information Technology |
Edirisinghe, Gayani Anuradha | Sri Lanka Institute of Information Technology |
Lokuliyana, Shashika | Sri Lana Institute of Information Technology |
Keywords: Social Implications of Technology, Signal and Image Processing, Biomedical Engineering
Abstract: This paper
presents a non-invasive method to detect Anemia (a low level of
Hemoglobin) easily. The Hemoglobin concentration in human blood is an
important substance to health condition determination. With the results
which are obtained from Hemoglobin test, a condition which is called as
Anemia can be revealed. Traditionally the Hemoglobin test is done using
blood samples which are taken using needles. The non-invasive Hemoglobin
measurement system, discussed in this paper, describes a better idea
about the hemoglobin concentration in the human blood. The images of the
finger- tip of the different hemoglobin level patients which are taken
using a camera is used to develop the neural network-based algorithm.
The pre-mentioned algorithm is used in the developed non-invasive device
to display the Hemoglobin level. Before doing the above procedure, an
account is created in the mobile app and a questionnaire is given to
answer by the patient. Finally, both the results which are obtained from
the mobile app and the device are run through a machine learning
algorithm to get the final output. According to the result patient would
be able to detect anemia at an early stage.
|
|
15:45-16:00, Paper ThC45.3 | |
A Rate Control Method for QoE Enhancement of TCP-Based Audiovisual and Haptic Interactive Communications |
|
Nunome, Toshiro | Nagoya Institute of Technology |
Ito, Atsunori | Nagoya Institute of Technology |
Keywords: Wireless Communications & Networks, Multimedia Engineering, Robotics, Control Systems & Theory
Abstract: This paper
proposes a control method for the transmission rate of haptic media to
avoid network congestion in audiovisual and haptic interactive
communications using TCP. The previous study has proposed a
media-adaptive buffering control for QoE enhancement of audiovisual and
haptic interactive communications over UDP. Most of the communications
on the Internet use TCP. When we communicate audiovisual and haptic
using TCP because of some restrictions, communication delay occurs by
such as retransmission control, and media output quality degrades. Then,
we control the media transmission rate according to network conditions
and enhance QoE. We employ an application-level QoS parameter in the
judgment of rate control of haptic media. We show that the interactive
communication of audiovisual and haptic using TCP is feasible as using
UDP through a subjective experiment.
|
| |