scispace - formally typeset
Search or ask a question
Topic

Sketch recognition

About: Sketch recognition is a research topic. Over the lifetime, 1611 publications have been published within this topic receiving 40284 citations.


Papers
More filters
Dissertation
01 Jan 2008
TL;DR: A speaker-independent continuous Hidden Markov hlodel (CHPllhI) algorithm for speech recognition is to be produced in the project, and the fixed point code is used instead of floating point code as the final CHAIN algorithm adopted would be ported to a selected DSP.
Abstract: The SpeechDat-Car database was built in the framework of the SpeechDat-Car project (European Union founded). The availability of the database allows the development of the speech recognition applications in mobile device under severe condition of car environment. A speaker-independent continuous Hidden Markov hlodel (CHPllhI) algorithm for speech recognition is to be produced in the project. To achieve this object ive, the CHh1M system adapt at ions and testings on the SpeechDat-Car database would be investigated. This is done by first understanding the algorithm and evaluating both the front end and back end of the system, and then, modifying the algorithm for obtaining better performance. As the final algorithm adopted would be ported to a selected DSP, as given in the system specification, the work also focuses on algorithms with low computational cost and memory consumption. The work is factored into two stages. In the first stage. one database (one language only) was processed and an optimum parameter set for the CHhIhI algorithm was derived. This derivation of optimum parameter set also involved the work on the model refinement and recognition algorithm enhancement. In the second stage, the algorithms used in the first stage was tested and extended to other languages. Throughout the project development, the fixed point code is used instead of floating point code as the final CHAIN algorithm adopted would be ported to a selected xii ATTENTION: The Singapore Copyright Act applies to the use of this document. Nanyang Technological University Library
Journal Article
TL;DR: Experimental results carried out prove that the combination of 2dimensional-DCT and artificial neural network tool used for face recognition improves the recognition rate significantly.
Abstract: Human faces carry huge and essential information which plays an important role in face detection and recognition Automatic recognition has various applications in diverse fields To recognize a face in real time is a challenging problem because pose variation is a major issue This paper presents a new robust technique for face recognition A two-dimensional discrete cosine transform which uses an image-based approach towards artificial is used for feature extraction The unwanted information is discarded and only the relevant data is extracted A neural network is used to train the database and also for recognition Two video sequences of a person are captured from a webcam mounted on a computer monitor Out of which one is for training phase and one is dedicated to testing phase The video sequences are taken under same illumination condition but with change in facial expression and pose variation Experimental results carried out prove that the combination of 2dimensional-DCT and artificial neural network tool used for face recognition improves the recognition rate significantly
Journal Article
TL;DR: An augmented new reality tool for vision-based hand gesture recognition in a camera-projector system and its main result is the quick on-line adaptation to the gestures of a new user to achieve user-independent gesture recognition.
Abstract: We developed an augmented new reality tool for vision-based hand gesture recognition in a camera-projector system. Our recognition method uses modified Fourier descriptors for the classification of static hand gestures. Hand segmentation is based on a background subtraction method, which is improved to handle background changes. Most of the recognition methods are trained and tested by the same service-person, and training phase occurs only preceding the interaction. However, there are numerous situations when several untrained users would like to use gestures for the interaction. In our new practical approach the correction of faulty detected gestures is done during the recognition itself. Our main result is the quick on-line adaptation to the gestures of a new user to achieve user-independent gesture recognition.
Journal Article
TL;DR: This work aims to find a quadratic algorithmic approach for the optimization of the temporary complexity of the Amazigh language recognition process, and also to maximize the rate of recognition of these characters facing some recognition problems.
Abstract: This paper describes an artificial neural network (ANN)-based system that uses a word frequency database for optical character recognition (OCR) of words in Amazigh - a North African language. OCR is a widely researched field with commercial systems available for common languages like English. In this research, we will study the character recognition using a new approach based on statistical calculation and artificial neural networks. What we propose in this work is a method that will exploit the power of the semantic analysis of the language to guess all the characters in order to solve the problem of similarity of characters and reduce recognition time. The approach we propose in this study will increase the recognition rate and optimize the response time. We have already several methods based on artificial neural networks, but most of these methods don't make an exact recognition with a rate of 100%, also the response time for these methods is very important. Optical character recognition (OCR) or video-coding means computer processes for translating text images printed or typed text files. It achieves far less than the human being, who, in addition to the recognition, understands the message, stores it or starts a critical analysis in a single time. And it's a very important topic area in the image processing science and automatic language processing. Several approaches and methods are invented to solve the problem of characters recognition, but the common problem among the majority of these approaches is response time, which is relatively important depending on the size of the instance and the recognition rate being usually lower than 100%. This work aims to find a quadratic algorithmic approach for the optimization of the temporary complexity of the Amazigh language recognition process, and also to maximize the rate of recognition of these characters facing some recognition problems. Many recognition methods have been proposed most of which is based on the treatment of individual character such as the extraction of the characteristics of each character. This type of method is very important although it does not meet 100% we need,

Network Information
Related Topics (5)
Feature (computer vision)
128.2K papers, 1.7M citations
84% related
Object detection
46.1K papers, 1.3M citations
83% related
Feature extraction
111.8K papers, 2.1M citations
82% related
Image segmentation
79.6K papers, 1.8M citations
81% related
Convolutional neural network
74.7K papers, 2M citations
80% related
Performance
Metrics
No. of papers in the topic in previous years
YearPapers
202326
202271
202130
202029
201946
201827