Code Number TITLE Abstract
SH19ML01 Experiments on Deep Face Recognition using Partial Faces

Abstract:- Face recognition is a very current subject of great interest in the area of visual computing. In the past, numerous face recognition and authentication approaches have been proposed, though the great majority of them use full frontal faces both for training machine learning algorithms and for measuring the recognition rates. In this paper, we discuss some novel experiments to test the performance of machine learning, especially the performance of deep learning, using partial faces as training and recognition cues. Thus, this study sharply differs from the common approaches of using the full face for recognition tasks. In particular, we study the rate of recognition subject to the various parts of the face such as the eyes, mouth, nose and the forehead. In this study, we use a convolutional neural network based architecture along with the pre-trained VGG-Face model to extract features for training. We then use two classifiers namely the cosine similarity and the linear support vector machine to test the recognition rates. We ran our experiments on the Brazilian FEI dataset consisting of 200 subjects. Our results show that the cheek of the face has the lowest recognition rate with 15% while the (top, bottom and right) half and the 3/4 of the face have near 100% recognition rates.

Keywords:- Face Recognition, Partial Face, Deep Learning, Cosine Similarity, Support Vector Machine.

SH19ML02 Predicting Diabetes in Healthy Population through Machine Learning

Abstract:- In this project, we revisit the data of the San Antonio Heart Study, and employ machine learning to predict the future development of type-2 diabetes.To build the prediction model,we use the support vector machines and ten features that are wellknown in the literature as strong predictors of future diabetes. Due to the unbalanced nature of the dataset in terms of the class labels, we use 10-fold cross-validation to train the model and a hold-out set to validate it. The results of this study show a validation accuracy of 84.1% with a recall rate of 81.1% averaged over 100 iterations. The outcomes of this study can help in identifying the population that is at high risk of developing type-2 diabetes in the future.

Keywords:- Disease Prediction,support vector machine,type 2 diabetes.

SH19ML03 Prediction of diabetic patient readmission using machine learning

Abstract:- Hospital readmissions pose additional costs and discomfort for the patient and their occurrences are indicative of deficient health service quality, hence efforts are generally made by medical professionals in order to prevent them. These endeavors are especially critical in the case of chronic conditions, such as diabetes. Recent developments in machine learning have been successful at predicting readmissions from the medical history of the diabetic patient. However, these approaches rely on a large number of clinical variables thereby requiring deep learning techniques. This article presents the application of simpler machine learning models achieving superior prediction performance while making computations more tractable

Keywords:- diabetes, hospital readmission, neural network, random forest, logistic regression.

SH19ML04 Wisture: Touch-less Hand Gesture Classification in Unmodified Smartphones Using Wi-Fi Signals

Abstract:- This project introduces Wisture, a new online machine learning solution for recognizing touch-less hand gestures on a smartphone (mobile device). Wisture relies on the standard Wi-Fi Received Signal Strength (RSS) measurements, Long Short-Term Memory (LSTM) Recurrent Neural Network (RNN) learning method, thresholding filters, and a traffic induction approach. Unlike other Wi-Fi based gesture recognition methods, theproposedmethoddoesnotrequireamodificationofthedevice hardware or the operating system and performs the gesture recognition without interfering with the normal operation of other smartphone applications. We discuss the characteristics of Wisture and conduct extensive experiments to compare the performance of the RNN learning method against state-of-theart machine learning solutions regarding both accuracy and efficiency. The experiments include a set of different scenarios with a change in spatial setup and network traffic between the smartphone and Wi-Fi access points (AP). The results show that Wisture achieves an online gesture recognition accuracy of up to 93% (average 78%) in detecting and classifying three gestures.

Keywords:- Wi-Fi, Radio Signal Strength, Gesture recognition, Mobile Phones, Machine Learning, Traffic Induction.

SH19ML05 Modeling Interaction Structure for Robot Imitation Learning of Human Social Behavior

Abstract:- This studypresents a learning-by-imitation technique that learns social robot interaction behaviors from natural human– human interaction data and requires minimum input from a designer. To solve the problem of responding to ambiguous human actions, a novel topic clustering algorithm based on action cooccurrence frequencies is introduced. The system learns human readable rules that dictate which action the robot should take, based on the most recent human action and the current estimated topic of conversation. The technique is demonstrated in a scenario where the robot learns to play the role of a travel agent. The proposed technique outperformed several baseline techniques in qualitative and quantitative evaluations. It responded more accurately to ambiguous questions and participants found it was easier to understand, provided more information, and required less effort to interact with.

Keywords:- Human–robot interaction, imitation learning, interaction structure, spoken dialog system, unsupervised learning.

SH19ML06 Automated Genre Classification of Books Using Machine Learning and Natural Language Processing

Abstract:- In today's world due to ever increasing demand to make computers perform tasks of humans, machine learning is used. It is a tedious task to manually read the entire book and classify it based on its genre. Novice writers find it troublesome to figure out the genre of their book, which can affect its reach to the right audience. The proposed method gains knowledge from a large number of words from the books and transforms them into a feature matrix. During transformation, the size of the initial matrix is reduced using Wordnet and Principle Component Analysis. Then, AdaBoost classifier is applied to predict the genres of the books.

Keywords:- Natural Language Processing, Machine Learning, Genre Classification, WordNet, TF-IDF, Decision Tree, AdaBoost, Principle Component Analysis (PCA).

SH19ML07 Computer-Vision-Based Surveillance of Intelligent Transportation Systems

Abstract:- This project focuses on the development of a video analytics processor for detection and classification of vehicles on motorways. The motivation for this project comes from the need to have a plug-and-play solution to analyse traffic, as most of the existing solutions require training some sort of structure to recognise objects on the scene. Such a module can work as a data input for traffic management systems in addition to more traditional sensors such as the magnetic loop detectors. We also present a novel approach to the vehicle classification problem based on the use of a fuzzy set. To illustrate the proposed approach, the detection and classification algorithms implemented were tested with different cameras in different scenarios, showing promising results.

Keywords:- computer vision, fuzzy set, object tracking, object classification, intelligent transportation systems.

SH19ML08 An Efficient Approach to Recognize Hand Gestures Using Machine-Learning Algorithms

Abstract:- Electromyography (EMG) from a subject’s upper limb can be used to train a machine-learning algorithm to classify different hand gestures. However, variability in the EMG signal due to between-subject differences can substantially degrade the machine-learning performance. This variation is usually due to the differences in both anatomical and physiological properties of the muscles, levels of muscle contraction, and inherent noises from the sensors. The aim of this study is to develop a subject-independent algorithm that can accurately classify different hand gestures. To minimize the between-subject differences, some selected time-domain EMG features are normalized to the area under the averaged root mean square curve (AUC-RMS). Five adult subjects with ages ranging 20-37 years performed three hand gestures including fist, wave-in, and wave-out for ten to twelve times each. Five machine-learning algorithms, including k-nearest neighbor (KNN), discriminant analysis (DA), Naïve Bayes (NB), Random Forest (RF), and Support Vector Machine (SVM) were used to classify the three different hand gestures. The results showed that the EMG features were moderately to strongly correlated with the AUC-RMS values. The SVM yielded maximum classification accuracy using the original EMG features (97.56%) which was significantly improved by using the normalized EMG features (98.73%) (p less than 0.05). The accuracy distribution of all classifiers were found to be closer to mean values when using the normalized EMG features compared to using the original EMG features. The developed approach of classifying different hand gestures will be useful in biomedical applications such as controlling exoskeletons and in certain human-computer interaction settings

Keywords:- hand gestures, electromyography, pattern recognition, machine learning.

SH19ML09 Mitch: A Machine Learning Approach to the Black-Box Detection of CSRF Vulnerabilities

Abstract:- Cross-Site Request Forgery (CSRF) is one of the oldest and simplest attacks on the Web, yet it is still effective on many websites and it can lead to severe consequences, such as economic losses and account takeovers. Unfortunately, tools and techniques proposed so far to identify CSRF vulnerabilities either need manual reviewing by human experts or assume the availability of the source code of the web application. In this paper we present Mitch, the first machine learning solution for the black-box detection of CSRF vulnerabilities. At the core of Mitch there is an automated detector of sensitive HTTP requests, i.e., requests which require protection against CSRF for security reasons. We trained the detector using supervised learning techniques on a dataset of 5,828 HTTP requests collected on popular websites, which we make available to other security researchers. Our solution outperforms existing detection heuristics proposed in the literature, allowing us to identify 35 new CSRF vulnerabilities on 20 major websites and 3 previously undetected CSRF vulnerabilities on 20 major websites and 3 previously undetected CSRF vulnerabilities on production software already analyzed using a state-of-the-art tool.

Keywords:- Cross-Site Request Forgery (CSRF); CSRF vulnerabilities.

SH19ML10 Mango Leaf Deficiency Detection Using Digital Image Processing and Machine Learning

Abstract:- Mango known to a national fruit of India, its leaves are exorbitantly affected by various nutrient deficiencies like nitrogen, iron, potassium and copper. These nutrients can change the natural color of mango leaves. Such leaves are considered to be deficient. The main purpose of this work is to detect various nutrient deficiencies of mango leaves. Initially a data set is created by extracting the different features of mango leaves using digital image processing. The extracted features include the RGB values and the texture of the leaves. This dataset is then used in the unsupervised machine learning model like clustering to cluster the various deficiencies which will help in further detection.

Keywords:- Machine Learning, Agriculture, K-Means Clustering, Image Processing, Deficiency.

SH19ML11 A New Approach For Vehicle Number Plate Detection

Abstract:- Identification of cars and their owners is a tedious and error prone job. The advent of automatic number plate detection can help tackle problems of parking and traffic control. The system is designed using image processing and machine learning. A new system is proposed to improve detection in low light and over exposure conditions. The image of vehicle is captured, which is preprocessed using techniques like grayscale, binarization. The resultant image is passed on for plate localization, for extracting the number plate using CCA (Connected Component Analysis) and ratio analysis. De-noising of number plate is done using various filters. The characters of the number plate are segmented by CCA and ratio analysis as well. Finally, the recognized characters are compared using techniques such as SVC (linear), SVC (poly), SVC (rbf), KNN, Extra Tree Classifier, LR+RF, and SVC+KNN. The proposed techniques help the system to detect well under dim light, over-exposed images and those in which the vehicle is angled.

Keywords:- Number plate detection, OCR , median filter , image processing , cca , image recognition, number recognition

SH19ML12 Detecting Ripe Canarium Ovatum (Pili) Using Adaboost Classifier and Color Analysis

Abstract:- Bicol region in terms of agriculture is known for its indigenous crop, canarium ovatum, or most commonly known as ‘Pili.’ Canarium Ovatum has been recognized for its economic importance due to its potential in the export market. However, there has been a growing demand for pili because of the lack of equipment in post-production and processing operations and the need of the market cannot be met by the growers and processors. Fruit Detection in harvesting one of the post-production processes, is the first major task of a robot. A vision system that can easily recognize fruit in a tree, which is levelled to be as intelligent as human beings is difficult to develop. This study helped increase the accuracy of the detection of ripe canarium ovatum. Images were captured using a high-end drone (Phantom 4 Professional with 20 megapixel resolution). This study established the data set by selecting images for training which is composed of 80% of the total image acquired and 20% for test set. The background information of the images like leaves, twigs, unripe pili, and other objects were also categorized. An Adaboost classifier and color analysis was used in the detection of ripe pili. An average of 90.77% accuracy of the ripe pili detection was recognized during the evaluation of the algorithm. The performance of the algorithm was evaluated according to true positive, false negative, and false positive with an average of 90.77%, 9.23% and 0.77% results, respectively. The detection algorithm achieved a high correct detection rate and the Haar-like features have potentials for extracting shape and texture information of the ripe pili in natural settings which contain various visual features due to complex structures of the leaves, twigs and other objects. Future research will include enhanced detection rates, reduced processing time, reduced manual processes, and various cultivated varieties of pili. It may also accommodate more varied unstructured environments.

Keywords:- canarium ovatum; adaboost; color analysis.

SH19ML13 Human Activity Recognition and Prediction Based on Wi-Fi Channel State Information and Machine Learning

Abstract:- At present, we are moving into the era of the Internet of Things. In this new era, it will be easy to find access points (APs) wherever we go. Signals from these APs can be used for more than just connecting to the Internet. The presence of a human between two APs and the human’s behavior cause a change in the waveform of a Wi-Fi signal. In this paper, we explain how changes in waveforms affect the channel state information of the signal and how machine learning can utilize that information to recognize and predict human behavior.

Keywords:- Channel State Information, Behavior Recognition, Internet of Things (IoT).

SH19ML14 A Mobile Application for Cat Detection and Breed Recognition Based on Deep Learning

Abstract:- Deep learning is one of the latest technologies in computer science. It allows using machines to solve problems in a manner similar to the human brain. Deep learning approaches have significantly improved the performance of visual recognition applications including image classification and image detection. In this paper we benchmark different deep learning models and present an Android application used to predict the location and breed of a given cat using a mobile phone camera. The average accuracy of the finalized model was 81.74%.

Keywords:- Deep Learning, Cat recognition, Android.

SH19ML15 Medical Imaging using Machine Learning and Deep Learning Algorithms: A Review

Abstract:- Machine and deep learning algorithms are rapidly growing in dynamic research of medical imaging. Currently, substantial efforts are developed for the enrichment of medical imaging applications using these algorithms to diagnose the errors in disease diagnostic systems which may result in extremely ambiguous medical treatments. Machine and deep learning algorithms are important ways in medical imaging to predict the symptoms of early disease. Deep learning techniques, in specific convolutional networks, have promptly developed a methodology of special for investigating medical images. It uses the supervised or unsupervised algorithms using some specific standard dataset to indicate the predictions. We survey image classification, object detection, pattern recognition, reasoning etc. concepts in medical imaging. These are used to improve the accuracy by extracting the meaningful patterns for the specific disease in medical imaging. These ways also indorse the decisionmaking procedure. The major aim of this survey is to highlight the machine learning and deep learning techniques used in medical images. We intended to provide an outline for researchers to know the existing techniques carried out for medical imaging, highlight the advantages and drawbacks of these algorithms, and to discuss the future directions. For the study of multi-dimensional medical data, machine and deep learning provide a commendable technique for creation of classification and automatic decision making. This paper provides a survey of medical imaging in the machine and deep learning methods to analyze distinctive diseases. It carries consideration concerning the suite of these algorithms which can be used for the investigation of diseases and automatic decisionmaking.

Keywords:- Medical imaging; Machine learning; Deep learning; Image enhancement; Information retrieval.

SH19ML16 Using Machine Learning for the detection of Radio Frequency Interference

Abstract:- Radio Astronomy, by its very nature, detects extremely faint cosmic radio signals and is therefore very susceptible to Radio Frequency Interference (RFI). We present some initial results of our work to identify RFI using a Machine Learning (ML) based approach. The data is taken from VLBI observations data from three well separated observatoriesinAustralia: ATCA,ParkesandMopra;andwework with the 2-bit data directly from the telescopes. Our approach uses a Generative Adversarial Network (GAN) and an autoencoder to perform unsupervised machine learning on the data.

Keywords:- Generative Adversarial Network (GAN); Radio Astronomy.


Abstract:- End-to-end Speech Translation (ST) models have many potential advantages when compared to the cascade of Automatic Speech Recognition (ASR) and text Machine Translation (MT) models, including lowered inference latency and the avoidance of error compounding. However, the quality of end-to-end ST is often limited by a paucity of training data, since it is difficult to collect large parallel corpora of speechandtranslatedtranscriptpairs. Previousstudieshaveproposed the use of pre-trained components and multi-task learning in order to benefit from weakly supervised training data, such as speech-totranscript or text-to-foreign-text pairs. In this paper, we demonstrate that using pre-trained MT or text-to-speech (TTS) synthesis models to convert weakly supervised data into speech-to-translation pairs for ST training can be more effective than multi-task learning. Furthermore, we demonstrate that a high quality end-to-end ST model can be trained using only weakly supervised datasets, and that synthetic data sourced from unlabeled monolingual text or speech can be used to improve performance. Finally, we discuss methods for avoiding overfitting to synthetic speech with a quantitative ablation study.

Keywords:- Speech translation,sequence-to-sequencemodel, weakly supervised learning, synthetic training data.

SH19ML18 Age Prediction based on Brain MRI Images using Extreme Learning Machine

Abstract:- Age Prediction, which means setting up a machine learning system, defined by using different sets of data for training, and then the estimation of the actual age of humans, is a subject that has been studied in recent years. To achieve this, researchers have been experimenting with various body components, such as DNA, speech signals, medical images, facial images, etc. Recent researches show that brain structure changes with age or psychiatric disorders. So a useful tool for estimating the age of humans is the brain's MRI images. Brain Magnetic Resonance Imaging (MRI) use radio waves and a robust magnetic field to create detailed images of the organs and tissues within the body. In this paper, the age of humans is predicted based on brain MRI images. To extract T1-MRI features, two different methods are proposed, then to estimate age, Extreme Learning Machine (ELM) is employed. Given that the amount of computations needed in this method and the time required to age estimation is low, the proposed method has acceptable performance.

Keywords:- MRI, Brain Age, Machine Learning.

SH19ML19 Design of Morse Message Transmission Control Algorithm Based on Machine Learning

Abstract:- The design of Morse code transmission control algorithm based on the machine learning are carried out due to the Morse message system that lacks flexibility and fails to meet the requirements of the reporting training at low and medium speeds. According to the characteristics of the report, the messages are classified into four categories, which all are designed report control algorithms or strategies, which solves the shortcomings caused by fixed bit rate and fixed ratio reporting in an effective way, and at the same time, offers a new idea for the overall design of the system.

Keywords:- machine learning; Morse code; control algorithm.

SH19ML20 Using Machine Learning to Represent Electromagnetic Characteristics of Arbitrarily-shaped Targets

Abstract:- A general data sparse representation of electromagnetic characteristics of an arbitrarily-shaped target is developed by using the machine learning model. The data sparse representation of the electromagnetic response is firstly figured out by the skeletonization technique. The machine learning approach is then employed to construct a general and flexible model which can capture the electromagnetic characteristics of the target of interest. Numerical experiments are conducted to validate the performance of the model.

Keywords:- Machine Learning, artificial neural network(ANN), Method of Moments(MoM), data sparse representation.

SH19ML21 Scene to Text Conversion and Pronunciation for Visually Impaired People

Abstract:- The recent technological advancements are focusing on developing smart systems to improve the quality of life. Machine learning algorithms and artificial intelligence are becoming elementary tools, which are used in the establishment of modern smart systems across the globe. In this context, an effective approach is suggested for automated text detection and recognition for the natural scenes. The incoming image is firstly enhanced by employing Contrast Limited Adaptive Histogram Equalization (CLAHE). Afterward, the text regions of the enhanced image are detected by employing the Maximally Stable External Regions (MESR) feature detector. The non-text MSERs are removed by employing appropriate filters. The remaining MSERs are grouped into words. The text recognition is performed by employing an Optical Character Recognition(OCR) function. The extracted text is pronounced by using a suitable speech synthesizer. The proposed system prototype is realized. The system functionality is verified with the help of an experimental setup. Results prove the concept and working principle of the devised system. It shows the potential of employing the suggested method for the development of modern devices for visually impaired people.

Keywords:- Image Processing, Text Recognition, MSER Features Detector, OCR, Speech synthesizer.

SH19ML22 Design of Intelligent Home control system based on Machine Learning

Abstract:- Aiming to improve the control ability of smart home, a design scheme of wireless communication control system for intelligent home based on machine learning is proposed. The wireless transmission module, the ZigBee data acquisition module, the man-machine interface module of the intelligent home control system are constructed. Machine learning algorithm is used to design the control system. The robustness of intelligent home control is improved by combining fuzzy PID control scheme. The steady-state error compensation method is adopted to improve the anti-dry ability of intelligent home control.

Keywords:- machine learning, smart home, control system design.

SH19ML23 EEG-Based Identification of Latent Emotional Disorder Using the Machine Learning Approach

Abstract:- Emotion influences our daily life to a large extent, especially for those who are undergoing bad mood and have high risk for emotional disorders. It is hard to recognize them, but very important so that we can provide intervention before them getting worse. This study used EEG signals to recognize who has high risk for emotional disorders instead of emotion type only. The proposed machine learning method combined the features of multiple cortex areas and frequency bands to find the high risky group for emotional disorders through a kernel SVM classifier. It achieved the accuracy of 95.20%, with all cortex areas and all frequency bands. Results showed that the frontal cortex, central cortex and temporal cortex have a primary influence on identifying emotional disorder and can be used for the reference information for professional diagnose.

Keywords:- emotional disorder; EEG; machine learning.

SH19ML24 Image Feature Recognition of Railway Truck Based on Machine Learning

Abstract:- Image feature is an important area of artificial intelligence. It refers to the technique of using computers to process, analyze, and understand images, aimed at recognizing targets and objects in various modes. Based on the machine learning algorithm, this paper uses the deep learning model to detect the truck number area under the complex background interfaces, and then uses the convolutional neural network to design the dynamic segmentation method for recognizing the truck number. The deep learning model faster regions with convolutional neural network(RCNN) can accurately detect the truck number area and make a good foundation for subsequent recognition.

Keywords:- Image feature recognition, Artificial intelligence, Machine learning, Convolutional neural network.

SH19ML25 Implementation of Machine Learning Algorithms for Autonomous Robot Trajectory Resolving

Abstract:- Machine learning in the modern industry and transport became a hot issue and its implementation finding broad field in recent activities in industry and daily life. Robotic is widely using machine learning of different type for image recognizing and for robotic trajectory generation in unknown environment. This paper provides overview of existing feature and tries to predict trend of machine learning development in the robotics. Use of connection between units and machine learning features can improve intelligence and capabilities of robots for trajectory generation. Reviewed cases provides nice results, available on the field of robotic trajectory and brings some breakthrough movement towards fully autonomous devices. Finally, discussion and conclusions are provided in the paper.

Keywords:- machine learning, trajectory generation, autonomous robotics, robot communication, robot teaching.

SH19ML26 Development of Android Application for Gender, Age and Face Recognition Using OpenCV

Abstract:- The idea behind the face recognition system is the fact that every individual has a unique face. Like the fingerprint, an individual's face has many unique structures and features. Facial authentication and facial recognition are challenging tasks. For facial recognition systems to be reliable, they must work with great precision and accuracy. Images captured taking into account different facial expressions or lighting conditions allow greater precision and accuracy of the system compared to a case where only one image of each individual is stored in the database. The face recognition method handles the captured image and compares it to the images stored in the database. If a matching template is found, an individual is identified. Otherwise, the person is reported as unidentified. This paper describes and explains in detail the entire process of developing Android mobile application for recognizing person’s gender, age and face. Face detection and recognition methods that have been used are described and explained as well as development tools used in the development of Android mobile application. The software solution describes the details of using the OpenCV library and shows the actual results of the mobile application through the images.

Keywords:- face detection; deep neural network; face recognition; Android application; gender, age, face recognition;OpenCV;Android Studio.

SH19ML27 Estimation of Speaker’s Confidence in Conversation Using Speech Information and Head Motion

Abstract:- Research and development of dialog robots for the purpose of conversation has been actively conducted in recent years. Many of these robot technologies are specialized in language information such as speech recognition and sentence comprehension. However, the majority of the information that human beings emphasize during dialog is nonverbal information. Even in the same words, the meaning of words changes as voice tones, expressions and gesture change. In order for a robot to realize a smooth dialog like a human being, it is important to combine language information and nonverbal information. Therefore, this paper focus on speech information and head motion to estimate the confidence of the user. The confidence conditions were defined as five-level status according to the questionnaire answer: most confident, comparative certainty, neutral, relative uncertain, most unconfident. Two experiments were conducted in this paper. The first experiment was conducted machine learning by summarizing all data of participants. The second experiment was conducted machine learning for each participant within creasing the number of data per participant. The best results could be classified as the most confident/unconfident with an accuracy of 91.3% analyzed by multilayer perceptron (MLP) via Weka. Accuracy increased by about 31% over the first experiment.This is considered to be the reason that the characteristics of individuals were manifested by doing machine learning with data of each participant.

Keywords:- multilayer perceptron (MLP);machine learning.

SH19ML28 Machine Learning Based Sleep-Status Discrimination Using a Motion Sensing Mattress

Abstract:- This project presents a novel sleep-status discrimination system by adopting a motion sensing mattress which detects the user’s activities on bed including the movement of head, chest, legs and feet. Unlike traditional methods like Polysomnography (PSG) which needs electrical equipment connected to users, or like wrist actigraphy which needs to be contact to users, the proposed system distinguishes sleep states in a non-conscious and non-contact way. The proposed system is built by a machine learning technique in the offline stage, and distinguishes sleep states in the online stage by using our designed sleep-status discrimination algorithm. The experimental results illustrate that the proposed method efficiently distinguishes sleep statuses without using a wearable device contact to body or using PSG diagnosis undertaken at hospitals.

Keywords:- Sleep-status discrimination; machine learning.

SH19ML29 An Opinion Mining For Indian Premier League Using Machine Learning Techniques

Abstract:- Social media has dramatically changed the way people express their opinion, appraisals or feelings towards entities or brand. Among many social media sites one of the free social service websites i.e. twitter, that permits users to publish their everyday life related events. As we know that twitter blog posts are being originated continually and Twitter having character short or limit to the Twitter posts (tweets) and also extremely compatible origins of continuous flow data for finding or detect opinion mining. Blog posts will reveal general or people emotions once taken in collection as an example throughout events like IPL 2016. Here our work presents, and provide the effectiveness of a machine learning model as positive or negative sentiment on tweets. The data collection of tweets and processing them by filtration best of the authorized IPL hastags (#IPL 2016 and #IPL 9)which can be done through using of Twitter’s API(Application Programming Interface) service. We analyze the performance of the ‘Random Forest’ against existing supervised machine learning algorithms with respect to its accuracy, specificity, sensitivity etc. here explanation of our paper is to performed opinion mining for the event like Indian Premier League 2016 effectively.

Keywords:- Social network, Opinion Mining, Machine Learning, Twitter, Stream Data Analysis.

SH19ML30 Floods Prediction Using Radial Basis Function (RBF) Based on Internet of Things (IoT)

Abstract:- Massive and continuously rainfall will cause floods. Floods can cause people's activities in the area to be hampered. With the technology that grows rapidly, people can get information easily. This Final Project is made to give information about the result of floods prediction using a technology called Internet of Things (IoT). This floods prediction is using Radial Basis Function. The data will be received from Citarum River Hall. The Information that used from Citarum River Hall is rainfall and river water debit. The result from Radial Basis Function Neural Network will be sent to an android application that displays the opportunity of flooding. Using epoch as much as 700 giving error value of TMA equal to 0.027 and error value of CH equal to 0.002, a learning rate of 0.00007 giving error value of TMA equal to 0.286 and error value CH equal to 0.002, and a hidden neuron of 2 giving error value of TMA equal to 0.6483 and error value of CH equal to 15.999 can be used to predict the flooding.

Keywords:- Radial Basis Function, Internet of Things, Floods Prediction.

SH19ML31 Demagnetization Modeling Research for Permanent Magnet in PMSLM Using Extreme Learning Machine

Abstract:- This project investigates the temperature demagnetization modeling method for permanent magnets (PM) in permanent magnet synchronous linear motor (PMSLM). First, the PM characteristics are presented, and finite element analysis (FEA) is conducted to show the magnetic distribution under different temperatures. Second, demagnetization degrees and remanence of the five PMs’ experiment sample are actually measured in stove at temperatures varying from room temperature to 300 °C, and to obtain the real data for next-step modeling. Third, machine learning algorithm called extreme learning machine (ELM) is introduced to map the nonlinear relationships between temperature and demagnetization characteristics of PM and build the demagnetization models. Finally, comparison experiments between linear modeling method, polynomial modeling method, and ELM can certify the effectiveness and advancement of this proposed method.

Keywords:- permanent magnets (PM), permanent magnet synchronous linear motor (PMSLM), extreme learning machine (ELM), linear modeling method, polynomial modeling method.

SH19ML32 Predicting Collaborative Performance at Assessment Level using Machine Learning

Abstract:- Most of the machine learning-based educational data mining (EDM) studies in university education merely focus on the predication of individual students’ performance at institutional/program and course levels. To predict collaborative performance, this study demonstrates the application of a rough set theory-based machine learning technique at the assessment level of a university course. It unveils if-then rules comprising key factors affecting assessment scores and categorizes them into performance classes of ‘Low’, ‘Medium’ and ‘High’. The results are applicable in chalking out strategies related to teaching and student advising to improve academic performance.

Keywords:- machine learning; collaborative learning; group performance prediction; educational data mining.

SH19ML33 Lane Detection and Tracking For Intelligent Vehicles: A Survey

Abstract:- Nowadays, Lane detection and tracking modules are considered as central requirements in every Intelligent Transportation System (ITS) development. The extracted lane information could be used in several smart applications for lane keeping systems, lane departure warning and avoiding collisions with other vehicles. In this proposed work, we are presenting, reviewing and comparing the different vision based algorithms used for detecting road lanes in autonomous vehicles.

Keywords:- Lane detection, Lane tracking, Machine Learning, Hough transform, Vision based algorithms, Image segmentation.

SH19ML34 Ship Detection Based on Deep Learning

Abstract:- China has abundant marine resources and a vast sea area. In military and civilian applications, the detection of targets for marine ships has important research significance. Complex and variable sea conditions make ship detection more difficult. In order to more accurately detect the ship's target, this paper proposes an improved YoloV3 algorithm to realize the endto-end ship target detection system. Our algorithm achieves highaccuracy fast ship identification by introducing CFE modules, improving loss functions, and augmenting data for small targets. The experimental results show that compared with the traditional machine learning target detection algorithm, our method has greatly improved the detection accuracy and detection rate. It achieves an accuracy of 74.8% and a detection rate of 29.8 frames. The ship detection system of this paper is excellent in detection accuracy and speed.

Keywords:- deep learning. Ship detection. complex background.

SH19ML35 A Novel Machine Learning Algorithm to ReducePrediction Error and Accelerate Learning Curve forVery Large Datasets

Abstract:- This paper presents a novel machine learning algorithm with an improved accuracy and a faster learning curve, for very large datasets. Previously, an algorithm using lr-partitions was designed to improve upon C4.5. However, this algorithm has a relatively high percentage of undefined combinations of attribute values in its final results, increasing the learning error. In this paper, a new type of clustering algorithm was proposed to generate output values for those undefined combinations, thus accelerating the learning curve and reducing the prediction error by several percentage points on various popular datasets from the UCI Machine Learning Database.

Keywords:- Machine Learning; Prediction Accuracy; Learning Curve; Multi-valued logic; Supervised Classifier.

SH19ML36 Network Intrusion Detection using SupervisedMachine Learning Technique with Feature Selection

Abstract:- A novel supervised machine learning system is developed to classify network traffic whether it is malicious or benign. To find the best model considering detection success rate, combination of supervised learning algorithm and feature selection method have been used. Through this study, it is found that Artificial Neural Network (ANN) based machine learning with wrapper feature selection outperform support vector machine (SVM) technique while classifying network traffic. To evaluate the performance, NSL-KDD dataset is used to classify network traffic using SVM and ANN supervised machine learning techniques. Comparative study shows that the proposed model is efficient than other existing models with respect to intrusion detection success rate.

Keywords:- intrusion, machine learning, deep learning,neural network, support vector machine, feature selection.

SH19ML37 Android Malware Detection Using Genetic Algorithm based Optimized Feature Selection and Machine Learning

Abstract:- Android platform due to open source characteristic and Google backing has the largest global market share. Being the world’s most popular operating system, it has drawn the attention of cyber criminals operating particularly through wide distribution of malicious applications. This paper proposes an effectual machine-learning based approach for Android Malware Detection making use of evolutionary Genetic algorithm for discriminatory feature selection. Selected features from Genetic algorithm are used to train machine learning classifiers and their capability in identification of Malware before and after feature selection is compared. The experimentation results validate that Genetic algorithm gives most optimized feature subset helping in reduction of feature dimension to less than half of the original feature-set. Classification accuracy of more than 94% is maintained post feature selection for the machine learning based classifiers, while working on much reduced feature dimension,thereby, having a positive impact on computational complexity of learning classifiers.

Keywords:- Android malware analysis; feature selection;Genetic algorithm; machine learning; reverse-engineering.

SH19ML38 Machine Learning Based Fault Type Identification In the Active Distribution Network

Abstract:- To realize the intelligent of the distribution network, it is necessary to identify the fault type accurately. This project presents the fault type identification method based on machine learning in active distribution networks. The process of machine learning is divided into four steps: data preparation,data pre-processing, feature extraction and model training. When preparing data, a method of generating fault scenarios in the batch of simulation experiments is presented. The IEEE34 Bus System is built in PSCAD to complete the data preparation for machine learning. Variation multiples of voltage and current are extracted as the features to describe the fault type. Various machine learning models are trained by cross-validation method to get the accuracy of identification. The application of decision tree in fault type identification is presented in the form of a tree diagram. The result of fault type identification is shown by the confusion matrix of the decision tree. All the test results show that the proposed fault identifiers can identify all kinds of fault types in the distribution network.

Keywords:- machine learning; fault type identification; active distribution network; batch simulation; feature extraction.

SH19ML39 Automatic Pedestrians Segmentation Based on Machine Learning in Surveillance Video

Abstract:- Pedestrian detection and segmentation play an important role in video surveillance. This paper presents a novel pipeline framework for automatic pedestrian detection and segmentation by combining machine learning with traditional computer visual methods. In particular, the Histogram of Oriented Gradient (HOG) and Support Vector Machine (SVM)are employed for pedestrian detection, and then the frame difference method is adopted for the tracking of the pedestrian. GrabCut and Mask R-CNN methods are used in the segmentation of pedestrians. The experiments are conducted on common benchmarks. The experimental results show that our method has made significant progress in automatic pedestrian detection and segmentation compared to the traditional Grabcut method.

Keywords:- Pedestrian detection; Histogram of Oriented Gradient; Support Vector Machine; GrabCut Segmentation Algorithm; Mask R-CNN.

SH19ML40 A Machine Learning Based Approach to Predict Power Efficiency of S-boxes

Abstract:- In the era of lightweight cryptography, designing cryptographically good and power efficient 4 × 4 S-boxes is a widely discussed problem. While the optimal cryptographic properties are easy to verify, it is not very straightforward to verify whether a S-box is power efficient or not. The traditional approach is to explicitly determine the dynamic power consumption using commercially available CAD tools and report accordingly based on a pre-defined threshold value. However, this procedure is highly time consuming, and the overhead becomes formidable while dealing with a set of S-boxes from a large space.This mandates development of an automation tool which should be able to quickly characterize the power efficiency from the Boolean function representation of an S-box. In this paper, we present a supervised machine learning (ML) assisted automated framework to resolve the problem for 4 × 4 S-boxes, which turns out to be approximately 14 times faster (using AND-OR-NOT gates) than the traditional approach. The key idea is to extrapolate the knowledge of the literal counts of various functional forms, AND-OR-NOT gate counts in the simplified SOP form of the underlying Boolean functions corresponding to the S-box to predict the dynamic power efficiency. We demonstrate the effectiveness of our framework by reporting a set of power efficient S-boxes from a large set of 4 × 4 optimal S-boxes. The experimental results and performance of our novel technique depicts its superiority with high efficiency and low time overhead.

Keywords:- Power Efficiency, Optimal S-box, Dynamic power, Machine Learning.

SH19ML41 PCB-METAL: A PCB Image Dataset for Advanced Computer Vision Machine Learning Component Analysis

Abstract:- We introduce PCB-METAL, a printed circuit board(PCB) high resolution image dataset that can be utilized for computer vision and machine learning based component analysis. The dataset consists of 984 high resolution images of 123 unique PCBs with bounding box annotations for ICs(5844), Capacitors(3175), Resistors(2670), and Inductors(542). The dataset is use-ful for image-based PCB analysis such as component detection, PCB classification, circuit design extraction,etc. We also provide baseline evaluations for IC detection and localization on state-of-the-art deep learning object detection algorithms.

Keywords:- printed circuit board(PCB); component detection.


Abstract:- We present a fingertip-shaped tactile sensor system that can measure static forces and slip vibrations using the same sensor. A fully integrated stress sensor ASIC leads to a simple design and assembly of the tactile fingertip.Instead of calibrating the fingertip to quantitatively measure forces, we use machine learning to extract abstract information out of the raw sensor data. Avoiding complex signal processing, this sensor-to-information processing scheme is fast and can have a small footprint. The results show that the system can classify the direction of applied forces with 99.8 % accuracy. The combination of the stress sensor array and the machine learning approach allows to detect slip and tangential force direction simultaneously.The combined classification achieves 99.6 % accuracy.

Keywords:- Tactile Sensor; Machine Learning; Slip Detection.

SH19ML43 Range-adaptive Impedance Matching of Wireless Power Transfer System Using a Machine Learning Strategy Based on Neural Networks

Abstract:- This work describes the implementation of a machine learning (ML) strategy based on the neural network for real-time range-adaptive automatic impedance matching of Wireless Power Transfer (WPT) applications. This approach for the effective prediction of the optimal parameters of the tunable matching network and classification range-adaptive transmitter coils (Tx) is introduced in this paper aiming to achieve an effective automatic impedance matching over a wide range of relative distances. We propose a WPT system consisting of a tunable matching circuit and 3 Tx coils which have different radius controlled by trained neural network models.The feed forward neural network algorithm was trained using 220 data and classifier’s in pattern recognition accuracy were characterized. The proposed approach achieves a Power transfer efficiency (PTE) around 90% for ranges within 10 to 25cm, is reported.

Keywords:- Impedance matching, machine learning, neural network, pattern recognition, resonant coupling, wireless power transfer.


Abstract:- In recent years, printable graphical codes have attracted a lot of attention enabling a link between the physical and digital worlds, which is of great interest for the IoT and brand protection applications. The security of printable codes in terms of their reproducibility by unauthorized parties or clonability is largely unexplored. In this paper, we try to investigate the clonability of printable graphical codes from a machine learning perspective. The proposed framework is based on a simple system composed of fully connected neural network layers. The results obtained on real codes printed by several printers demonstrate a possibility to accurately estimate digital codes from their printed counterparts in certain cases. This provides a new insight on scenarios, where printable graphical codes can be accurately cloned.

Keywords:- Printable graphical codes, clonability at-tack, machine learning.

SH19ML45 A Machine Learning Approach for Heart Rate Estimation from PPG Signal using Random Forest Regression Algorithm

Abstract:- In this paper, a new method is proposed to estimate the heart rate (HR) from wearable devices. A promising feature in today's world of HR monitoring is Photoplethysmography (PPG). However, during physical exercise HR estimation accuracy is seriously affected by noise and motion artifacts (MA). To reduce the effect of MA and estimate HR changes there are many conventional algorithms.Here, a new approach to estimate HR which is called multi-model machine learning approach (MMMLA) is shown. In this proposed algorithm, it firstly trains and tests the model for the different feature and different data set. Then it separates noisy and non-noisy data by K-means clustering. This lets the machine learn from noisy and non-noisy data. Then the Random Forest Regression algorithm is used to fit data and predict HR from test data. Here, feature engineering is also done, in other words, a different set of the feature is chosen and check their behavior with our proposed model and the error rate for every set of the feature was calculated. The mean absolute error and root mean square (RMS) error of HR was calculated. The lowest mean absolute error found in this research was 1.11 beats per minute (BPM). This result shows the capability of proposed machine learning-empowered system in HR estimation from PPG signal.

Keywords:- Heart Rate, Photoplethysmography, Signal processing, Machine learning, Feature engineering.

SH19ML46 Breast Cancer Diagnosis Using Image Processing and Machine Learning for Elastography Images

Abstract:- As a trending medical imaging technique, Elastography and B-mode (ultrasound) are combined as a diagnostic tool to differentiate between benign and malignant breast lesions based on their stiffness and geometric properties.Image processing techniques are applied to the resulting images for feature extraction. Data pre-processing methods and principal component analysis (PCA) as a dimensionality reduction technique are applied to the dataset. In this paper, supervised learning algorithm “support vector machine (SVM)” is used for the classification of combined elastogram and B-mode images.Model validation is performed with K-fold cross-validation to ensure the generalization of the algorithm. Accuracy, confusion matrix, and logistic loss are then evaluated for the used algorithm. The maximum classification accuracy is 94.12% when using SVM with radial basis function (RBF) kernel.

Keywords:- Breast Cancer, Elastography, Image Processing,Principle component analysis, Support Vector Machine (SVM).

SH19ML47 Generate use case from the requirements written in a natural language using machine learning

Abstract:- Recently it has become important to focus on the requirements of the system and how to take them and analyze them to determine the system infrastructure through which they will be relied upon in the rest of the system building. Some difficulties have been encountered in the process of understanding and analyzing the data taken from the user to convert it to UML diagrams. In this paper, we create a new approach that focuses mainly on increasing accuracy for this technique, reducing time in the systems of generating the use case of text written in natural language and finding solutions to some problems in current technologies because people need a smart and accurate system to meet their needs and save their time and increase the reliability of the reliance on software.

Keywords:- Use Case Diagram, Machine Learning, Natural Language Processing, User requirements analysis, Automatic Diagrams Generation.

SH19ML48 Plant Disease Classification Using SOFT COMPUTING Supervised Machine Learning

Abstract:- Plants are always concerned about the diseases introduced by pathogens For example infections,microorganisms and parasites in the plant bodies.It is globally recognized that, pathogens tends to cause huge yield misfortunes. Various researchers have explored how to diminish the harmfulness of pathogens in plants. A few analysts have explored some opposition qualities in plants and attempts to improve the obstruction of plants to pathogens.Meanwhile, different analysts have created ID and scoring framework for monitoring and examining the advancement or quality and also by anticipating the infection bolstered leaves. The reason for this Review work is to display the use of AI in the revelation of plant opposition.

Keywords:- Machine learning; plant diseases algorithm, Duplicate purchase, Separation and feature removal.

SH19ML49 Early Detection System for Gas Leakage and Fire in Smart Home Using Machine Learning

Abstract:- Making houses more inclusive, safer, resilient and sustainable is an important requirement that must be achieved in every society. Gas leakage and fires in smart houses are serious issues that are causing people’s death and properties losses. Currently, preventing and alerting systems are widely available. However, they are generally individual units having elementary functions without adequate capabilities of multi-sensing and interaction with the existing Machine-to-Machine (M2M) home network along with the outside networks such as Internet. Indeed,this communication paradigm will be clearly the most dominant in the near future for M2M home networks. In this paper, we are proposing an efficient system model to integrate the gas leakage and fire detection system into a centralized M2M home network using low cost devices. Then, through machine learning approach,we are involving a data mining method with the sensed information and detect the abnormal air state changes in hidden patterns for early prediction of the risk incidences. This work will help to enhance safety and protect property in smart houses.

Keywords:- smart home, gas leakage detection, fire detection,machine-to-machine, wireless sensor network, machine learning.

SH19ML50 Design of a Biosignal Based Stress Detection System Using Machine Learning Techniques

Abstract:- This study represents a design of a detection system of stress through machine learning using some available bio signals in human body. Stress can be commonly defined as the disturbance in psychological equilibrium. Stress detection is one of the major research areas in biomedical engineering as proper detection of stress can conveniently prevent many psychological and physiological problems like cardiac rhythm abnormalities or arrhythmia. There are several bio-signals available (i.e. ECG, EMG, Respiration, GSR etc.) which are helpful in detecting stress levels as these signals shows characteristic changes with stress induction. In this paper,ECG was selected as the primary candidate because of the easily available recording (i.e. several mobile clinical grade recorders are available now in the market) and ECG feature extraction techniques. Another advantage of ECG is that respiratory signal information can also be detected form ECG which is known as EDR (ECG derived Respiration) without having separate sensor system for respiration measurement.Features of ECG signals are distinctive and collection of the signals is cost-efficient. From ECG we derived RR interval, QT interval, and EDR features for the development of the model.For the implementation of a supervised machine learning(SVM) method in MATLAB, Physionet’s “drivedb” database was used as the training dataset and validation. SVM was chosen for classification, as there are two classes of labeled data; ‘stressed’ or ‘non-stressed’. Several SVM model types were verified by changing the feature number and Kernel type.Our results showed an accuracy level of 98.6% with Gaussian Kernel function and using all available features (RR, QT and EDR), which also emphasizes the importance of respiratory information in stress detection through Machine Learning.

Keywords:- stress detection, arrhythmia, ECG Derived Respiration (EDR), Machine Learning, MATLAB.

SH19ML51 The Color-Music Relationship Direct Modeling using Machine Learning

Abstract:- As a method to improve the effectiveness of music retrieval, this paper suggests a method that can exclude subjective judgment in music retrieval by directly modeling the color-music relationship and can assure perpetuity by automating color discernment on new music. To this end, a method to define representative color of each music from colors that people reminisce after hearing music is presented and color-music model that classifies feature features of music by using machine learning is also presented.The experiment shows performance improvement by 4% compared with VGG-19.

Keywords:- Color-Music Relationship; Music Classification;Music Retrieval; Convolutional Neural Network; Deep Learning.

SH18ML52 Handwritten Character Recognition Using Deep-Learning

Abstract: This project seeks to classify an individual handwritten word so that handwritten text can be translated to a digital form. We used two main approaches to accomplish this task: classifying words directly and character segmentation. For the former, we use Convolutional Neural Network (CNN) with various architectures to train a model that can accurately classify words. For the latter, we use Long Short Term Memory networks (LSTM) with convolution to construct bounding boxes for each character. We then pass the segmented characters to a CNN for classification, and then reconstruct each word according to the results of classification and segmentation.

SH18ML53 Unsupervised machine learning for clustering the infected leaves based on the leaf-colours

Abstract: In data mining, the clustering is one of the important process for categorizing the elements into groups whose associated members are similar in their features. In this paper, the plant leaves are grouped based on the colors in the leaves. Totally, three categories are specified to represent the leaf with more green, leaf with yellowish shades and leaf with reddish shades. The task is performed using image processing. The leaf images are processed in the sequence such as image pre-processing, segmentation, feature extraction and clustering. Pre-processing is done to denoize, enhance and background color fixing for betterment of result. Then, the color-based segmentation is done on the pre-processed image for generating the sub-images by clustering the pixels based on the colors. Next, the basic features such as entropy, mean and stadard deviation are extracted from each sub-images. The extracted features are used for clustering the images based on the colors. The image clustering is done by the Neural Network architecture, self-organizing map (SOM) and K-means algorithm. They are evaluated with various distance measuring functions. Finally, the city-block in both method produced the clusters with same size. This cluster set can be used as a training set for the leaf classification in future.

SH18ML54 Real time license plate detection using openCV and tesseract

Abstract: This paper presents the implementation of image to text conversion. The paper describes various steps required to extract text from any image file (jpeg/png) and create a separate text file consisting of information extracted from image file. It considers the shortcomings of various image processing applications available and works on overcoming them by employing variable level of image processing and filtration. The CV2 OpenCV library using Python language is used for image processing and Tessaract is used for text extraction from the processed image. The variable level of image processing ensures that different images get different levels of treatment in order to produce optimized text results. After the image processing step is employed the output text file are formatted by filtering out commas, semicolons, apostrophes, colons and other such characters using ASCII filtering as these characters are not part of any standard license plate.

SH18ML55 Smart home automation with a unique door monitoring system for old age people using Python, OpenCV, Android and Raspberry pi

Abstract: In this paper, smart home automation system particularly for old age people is proposed based on python, OpenCV, raspberry pi and android application. The appliances are controlled by the Raspberry pi server, which operates according to the user command (touch or voice) received from the mobile phone. A unique door monitoring system is designed based on face detection and recognition from a camera installed outside the main door, which can be accessed from the phone using android application. One interesting feature that has been added is that, all the appliances can also be controlled through the voice of user. For energy efficiency user can analyze the usage of each appliance from their phone. Moreover, user can also control the intensity of light as well as the speed of the fan. With all this features incorporated in a single system with good and simple user interface, this system is cost effective and perfect for old age people living alone in their houses.

SH18ML56 Real Time Object Detection and Tracking Using Deep Learning and OpenCV

Abstract: Deep learning has gained a tremendous influence on how the world is adapting to Artificial Intelligence since past few years. Some of the popular object detection algorithms are Region-based Convolutional Neural Networks (RCNN), Faster-RCNN, Single Shot Detector (SSD) and You Only Look Once (YOLO). Amongst these, Faster-RCNN and SSD have better accuracy, while YOLO performs better when speed is given preference over accuracy. Deep learning combines SSD and Mobile Nets to perform efficient implementation of detection and tracking. This algorithm performs efficient object detection while not compromising on the performance.

SH18ML57 Design of plant disease detection system: A transfer learning approach work in progress

Abstract: The use of ICT has become essential to help farmers collect important and updated information and knowledge which are valuable resources that farming depends on. The study embarked investigates the problem sources of unavailability or lack of timely, relevant and accurate farming information and knowledge for small-scale farmers. The main target is to deal with plant diseases and how to manage them by carefully diagnosing the plants leaves. This work proposes to use image analysis and convolutional neural networks and the ever increasing capability of machine learning such as supervised learning to offer a mobile solution. A Design Science Research Methodology was followed in shaping skeleton of the proposed prototype. The developed prototype will be subjected to three usability measures to test if indeed it is timely, relevant and accurate as the farmers need it to be

SH18ML58 Automatic Vehicle License Plate Recognition System for Smart Transportation

Abstract: Metropolitan cities around the world are putting effort towards a common goal of smart and sustainable urban development as smart cities initiative. To realize smart transportation systems as part of smart cities, automatic recognition of vehicle license plates is essential for border checkpoint control, traffic and red-light violation, monitoring of vehicles entering and leaving critical infrastructures and government agencies. This paper presents an implementation of automatic license plate recognition system using vehicle license plates in Myanmar as a case study. The proposed approach can be used to train for recognition of country-specific vehicle license plates. Collecting more than 1200 actual license plate images, training and evaluating the performance, our implementation achieves 90% accuracy for recognizing characters of the license plates, and 100% accuracy for detecting total number of vehicle license plates in the videos. To the best of our knowledge, we are the first in Myanmar having a dataset of actual license plate images/videos and successfully implemented such an automatic recognition system.

SH18ML59 Facial emotion recognition in real-time and static images

Abstract: Facial expressions are a form of nonverbal communication. Various studies have been done for the classification of these facial expressions. There is strong evidence for the universal facial expressions of eight emotions which include: neutral happy, sadness, anger, contempt, disgust, fear, and surprise. So it is very important to detect these emotions on the face as it has wide applications in the field of Computer Vision and Artificial Intelligence. These fields are researching on the facial emotions to get the sentiments of the humans automatically. In Robotics, emotions classification can be used to enhance human-robot interactions since the robot is capable of interpreting a human reaction [13]. In this paper, the emotion detection has been done in both real-time and static images. In this project, we have used the Cohn-Kanade Database (CK) and the Extended Cohn-Kanade (CK+) database, which comprises many static images 640 × 400 pixels and for the real-time using the webcam. The target expression for each sequence in the datasets are fully FACS (Facial action coding system) coded and emotion labels have been revised and validated [3]. So for emotion recognition initially we need to detect the faces by using HAAR filter from OpenCV in the static images or in the real-time videos. Once the face is detected it can be cropped and processed for further detection of facial landmarks. Then using facial landmarks the datasets are trained using the machine learning algorithm (Support Vector Machine) and then classified according to the eight emotions [1]. Using SVM we were getting the accuracy of around 93.7%. These facial landmarks can be modified for getting better accuracy.

SH18ML60 Raspberry Pi and computers-based face detection and recognition system

Abstract: This paper aims to deploy a network that consists a group of computers connected with a microcomputer with a camera. The system takes images of people, analyze, detect and recognize human faces using image processing algorithms. The system can serve as a security system in public places like Malls, Universities, and airports. It can detect and recognize a human face in different situations and scenarios. This system implements “Boosted Cascade of Simple Features algorithm” to detect human faces. “Local Binary Pattern algorithm” to recognize these faces. Raspberry Pi is the main component connected to a camera for image capturing. All needed programs were written in python. Tests and performance analysis were done to verify the efficiency of this systemi

SH18ML61 Real Time Somnolence Detection System In OpenCV Environment for Drivers

Abstract: Safety is the first importance during driving. One fault of the driver can lead to severe physical harms, deaths and major economic losses. Today's, So many systems presented in automotive industries like navigation, different sensors, anti-breaking system, Stability control system etc. to make driver's effort easy. There are numerous reasons especially human mistakes which gives increase to road accidents. National Crime Bureau's reports say that there is an enormous increase in the road accidents in our country since last some years. The one of the reason happening the public road accidents are drowsiness of driver during driving. It is essential step to come with an effective method to detect somnolence as soon as driver senses drowsy. This could save huge number of accidents to happen. We made the system which Detect the Driver is sleeping or not, while driving the car or any Giant Vehicle, if driver is detected sleepy then system will warn the driver to wake up or take nap or offer coffee, this system will help to reduce the accident.

SH18ML62 Development of road sign recognition for ADAS using OpenCV

Abstract: Real time Road sign recognition technology of advanced driver assistance systems (ADAS) provide necessary information and instructions to help the driver to drive safely. Road sign recognition is the technology of driver assistance system which interprets the signs to the driver. Recognition is dependent on the combination of detection and classification. Among the various available methods the most efficient one is chosen. Thus detection of region of interest is performed by using Histogram of oriented gradient and classification by using support vector machine. Training data is generated from our own database. This paper represents a study to recognize road signs using OpenCv techniques. This is implemented in visual studio and ported on NVIDIA's TK1 platform. The experimental results shows good performance for recognition of ideogram based signs with an average speed of 25 frames per second having accuracy up to 94%.

SH18ML63 The Implementation of Lane Detective Based on OpenCV

Abstract: In intelligent vehicle systems, lane detection is one of the most important parts. This paper presents a lane detection algorithm that based on Hough transform. Principle of the algorithm and the implementation base on OpenCV are discussed in detail. The algorithm was verified at the end of this paper.

SH18ML64 Vehicle detection and counting using haar feature-based classifier

Abstract: In this paper we would describe a vehicle detection technique that can be used for traffic surveillance systems. An intelligent traffic surveillance system, equipped with electronic devices, works by communicating with moving vehicles about traffic conditions, monitor rules and regulations and avoid collision between cars. Therefore the first step in this process is the detection of cars. The system uses Haar like features for vehicle detection, which is generally used for face detection. Haar feature-based cascade classifiers are an effective object detection method first proposed by Viola and Jones. It's a machine learning based technique which uses a set of positive and negative images for training purpose. Results show this method is quite fast and effective in detecting cars in real time CCTV footages

SH18ML65 Enhanced smart doorbell system based on face recognition

Abstract: In recent years considerable progress has been made in the area of face recognition. Through the work of computer science engineers, computers can now outperform humans in many face recognition tasks, particularly those in which large databases of faces must be searched. A system with the ability to detect and recognize faces has many potential applications including crowd and airport surveillance, private security and improved human-computer interaction. An automatic face recognition system is perfectly suited to fix security issues and offer flexibility to smart house control. This project aims to replace costly image processing boards using Raspberry pi board with ARMv7 Cortex-A7 as the core within Opencv library. This project is mainly based on image processing by porting the Opencv library to the Raspberry Pi board. Algorithm for face recognition, based on principal component analysis (PCA), is programmed and implemented on the platform. The system is based on the criteria of low power consumption, resources optimization, and improved operation speed. This paper reviews the related work in the field of home automation systems and presents the system design, software algorithm, implementation and results.