54 research outputs found
Extreme learning machine based speaker verification
No full textSpeaker recognition is a technique of automatically recognizing who is speaking by using the speaker-specific information. Speaker verification is one of the speaker recognition aspects. It is the process of accepting or rejecting the identity claimed by a speaker. Extreme Learning Machine (ELM) is one of the computational intelligence techniques. It was proposed by Prof. Huang Guangbin, School of Electrical & Electronic Engineering, Nanyang Technological University. The past research has shown that ELM is superior to other techniques in the speech processing aspect. So it is worth to apply ELM to the speaker verification system that will make the system more efficient and reliable. The aim of this project is to develop a text-independent speaker verification system base on ELM. This report will show how the author achieves this goal. It includes the basic principle for speaker verification system, selection of speech corpus, feature extraction technique, speaker modeling technique and programming process etc. The author also provides detailed process of the project implementation, and some discussion on certain issues found in the project. At last, the author will provide recommendations for future work about this project.Bachelor of Engineerin
Human hand sign language recognition based on extreme learning machine
No full textAs machine learning algorithms and computer processing speed greatly advanced in recent years, real-time hand gesture recognition has become a promising topic in computer science and language technology. Some of the existing limits in achieving user-friendly experience are real-time recognition speed and accuracy. This project aims to realize practical dual hand real-time recognition and to develop new man-machine interaction functions. Based on senior Mr. Jiang Runzhou’s FYP work, Ms. Cai Xiao, Mr. Liu Hongyang and the author work closely to achieve the objective. Realizable functions include PowerPoint slide show control, music player control and Rock, Paper, Scissor game. Mr. Jiang Runzhou’s past gesture recognition is also enhanced to achieve more excellent accuracy.Bachelor of Engineerin
Extreme learning machines based human hand sign language recognition
No full textIn the field of Human Machine Interaction, with the development of 3D cameras and computer vision, recognition of human hand gestures and motions has become a real trending topic and is gaining greater significance in Natural User Interface (NUI). By detecting, tracking and recognizing hand gestures and motions, together with the development of graphical user interfaces, the use of traditional input devices such as key boards and mouse could be reduced. However, challenges have been faced as the accuracy and speed of real-time gesture recognition are seen as one of the bottlenecks with the current development. The final year project, Extreme Learning Machines based Human Hand Sign Language Recognition, is hence proposed with the aim of investigating real-time recognition of three broad classifications of gestures – static, dynamic, and motion detection with respective applications. Previous studies and available algorithms are examined to define the gap between the performance of Extreme Learning Machines proposed by Prof. Huang and other kernel methods. Suitable development platform, techniques for feature extraction and refinement, application of ELM methods are selected for the research. The desktop, game, and PowerPoint slides control application is developed based on 3Gear Nimble hand tracking library. Three applications for static gesture recognition, namely simultaneous detection of both hands, music player triggering application, and rock paper scissors game application, are presented. The dynamic gesture recognition is to combine both training and normal operation together for four main gestures. Windows default music player could also be triggered by making use of two of the dynamic gestures. The motion detection part of this project achieves recognition of writing number from 0 to 9 based on a countdown system, however, receiving not very nice performance. Performance of these three broad ranges of gesture recognition applications is evaluated compare with support vector machines. Recommendations for future work regarding the project is provided, in terms of achieving real-time recognition and performance improvement.Bachelor of Engineerin
Extreme learning machine based tracking-learning-detection framework in surveillance video
No full textThe importance of surveillance videos is gaining more attention from companies and organization nowadays. As one of the important methodologies in surveillance video and computer vision domain, object tracking emerged as a trending topic. Numerous researches have been done to further explore this topic. In this project of seeking object tracking algorithm optimisation, the author went through the full process of the research including data collection, literature review, implementation, evaluation and modification.
In the data collection part, the author compared different datasets including public datasets, sample videos from Delta Company and YouTube videos. After thorough examination, dataset “visual tracking benchmark” has been selected for the later evaluation process.
In the implementation process part, the author got a detailed research and understanding of the algorithm and source code for the TLD framework. Detailed comments for the most important two files are illustrated in the report.
In the evaluation part, visual tracking benchmark datasets are used for evaluation including two indicators: numbers of the successfully tracked frames and average localisation errors. Both advantages and shortcomings are identified in this section.
In the modification part, based on the secondary research, the author implemented an ELM (Extreme Learning Machine) based TLD framework which replaced the original k-NN classifier with extreme learning machine classifier to achieve better classification accuracy and smaller standard derivation. Differences in performance of the original and new algorithms are evaluated using the two indicators mentioned above.
Lastly, some head-ups are recommended by the author as a guide for future researchers to improve the current algorithm and implementation.Bachelor of Engineerin
Modelling temporal contextual information in eye movement data with application to gaze gesture recognition
No full textIn recent 20 years, technology has expanded on in-car human machine interaction (HMI). However, driver distraction has become a growing safety concern. Scientists try to construct systems to detect driver’s state to prevent driver distraction by tracking driver’s eye movements. Traditionally, eye data, such as gaze position, fixations or saccades are usually used as features in monitoring driver’s state. A very robust method is to use temporal contextual information, which is extracted from scan path and can keep more eye movement information. However, there is lack of systematic research into different ways of modelling temporal contextual information in eye movement data. Therefore, the author investigates three methods of modelling temporal contextual information. And to have a better understanding, the author uses the application of eye gaze gesture recognition to compare the methods and algorithms. Furthermore, the author implemented the application of gaze gesture recognition as a pilot research to examine if it is possible to apply in driving. As a result, the author provides insights on different methods and also the application itself can also serve as a prototype for further driving related applications.Bachelor of Engineerin
Machine learning based characters recognition
No full textThis final year project aims to study and implement some machine learning techniques for character recognition. The author was tasked to develop a mobile app for a business card scanner based on these techniques. The author has chosen to do research on Tesseract, which is an open-source optical character recognition (OCR) engine sponsored by Google and has embedded the Tess-two library locally into the business card scanner. The scanner was developed for Android systems. It is able to scan characters on business cards, distinguish the information and save it into the entry attributes for a new contact. It includes functions of photo cropping and saving, character recognition, information extraction and contact adding. The app design, app structure, key codes and testing results will be included in this report. Since OCR is the key technology of the application, its principle and development will be discussed for basic understanding as well as future improvement of the scanner.Bachelor of Engineerin
Industrial attachment report with ExxonMobile Asia Pacific Pte Ltd
No full textThe 22-weeks (5rd January to 6th June 2005) Industrial Attachment (IA) program provided the author an opportunity to apply what she had learnt from her course of study into practical application under Distributed Control System/Triple Modular Redundancy ( DCS/TMR) section, Plant Computing Group (PCG) of ExxonMobil Asia Pacific Pte Ltd. This report aims to provide a detailed summary on the experiences gained by the author and her applications during her attachment period in ExxonMobil Asia Pacific Pte Ltd. The report described in details the major project she had handled, which included the introduction, analyzing, developing, testing, implementing, problems encountered and their solutions, and also the recommendations of the author. The author had improved her analyzing, testing, problem solving skills and also communication ability after going through this period of attachment. This attachment has prepared the author for the real industry, giving her a head start for her career after her graduation
Effects of superheated steam on the tensile and interfacial adhesion properties of recycled carbon fiber
No full textBuilding and object detection for a smart domestic robot
No full textSmart domestic use robot has been a booming industry recently. The project group aimed to build a smart robot, which is able to carry out mapping, navigation, voice and speech recognition, gesture detection, and face and motion detection as well. As the one in charge of building the chassis and hardware implementation, the author will introduce the procedures of designing, building and programming the chassis. The procedures to grant the chassis with the capability to sense and locate obstacle, and the capability to perform navigated and non-navigated movements.Bachelor of Engineerin
EEG based assisted driving system
No full textThe EEG technologies have further developments in recent years. The use of EEG is not limited to traditional clinic uses. EEG-based applications become popular topics for the researchers and developers. This project develops an EEG-based assisted driving system to achieve using brainwave to control a car. It is an innovation for modern cars. The system translates subject’s brainwave to control commands of turning left, turning right, moving forward and moving reverse. It uses smooth to preprocess the signal, four methods including Sample Entropy, Power Spectrum Density, Spectrogram, and Continuous Wavelet Transform to extract EEG features, and Extreme Learning Machine to classify these features. The system achieves an accuracy of 0.9691 with an online dataset and its success is verified with EEG signals collected by the author via Emotiv.Bachelor of Engineerin
- …
