Browsing by Author "Kumarika, B.M.T."

Now showing 1 - 3 of 3

Deep learned Visual Model for Human Computer Interaction (HCI)
(Staff Development Center, University of Kelaniya, Sri Lanka, 2015) Kumarika, B.M.T.; Dias, N.G.J.
Background and rationale: The modern hand gesture recognition approaches can be classified as ‘contact’ and ‘vision’ based. Contact based approaches like Data Glove require a physical contact which can cause health issues and uncomfortable for some users. In contrast, users wear nothing in vision-based approaches where camera(s) capture the images of hands interacting with computers (Dan & Mohod, 2014).Therefore, vision-based approach is simple, natural and convenient. However, challenges to be addressed include illumination change, varying sizes of hand gestures, background clutters in visual patternidentification(Symonidis K,2000). Aim:Therefore, thepractical applicationof computer vision-based hand gesture recognition systems necessitates an efficient algorithm capable ofhandling those challenges. Theoretic al underpinning / Conceptual framework: As a solution to the complexity of the problem this research proposes Deep Neural Network (DNN) as robust, deep learned visual model. Deep learning attempts to model high-level abstractions (features) in data by using a biologically inspired model. In deep learning, the visual cortex of our brain is well-studied and shows a sequence of areas each of which contains a representation of the input, and signals flow from one to the next. Thus, each level of this feature hierarchy represents the input at a different level of abstraction, with more abstract features further up in the hierarchy, defined in terms of the lower-level ones where classification will be easy. Proposedmethodology:Created database of the hand gesture images is used for training and testing. Greedy layer-wise training is used to avoid the problems of training deep net in supervised fashion such as slow training, over fitting and unlabelled data. The results will be compared with test data which is a 15% of the data set.The results of the two tests oftraditional networksand deep network willalsobecompared. Expected outcomes: This will provide a robust Deep Neural Network as an efficient visual pattern recognition algorithm for real time hand gesture recognition.
Deep Unsupervised Pre-trained Neural Network for Human Gesture Recognition
(Faculty of Graduate Studies, University of Kelaniya, 2015) Kumarika, B.M.T.; Dias, N.G.J.
Recognition of visual patterns for real world applications is a complex process that involves many issues. Varying and complex backgrounds, bad lighting environments, person independent gesture recognition and the computational costs are some of the issues in this process. Since human gestures are perceived through vision, it is a subject of visual pattern recognition. Hand gesture recognition is of higher interest for Human-Computer Interaction (HCI), due to its widespread applications in virtual reality, sign language recognition, robot control, medical industry and computer games. The main goal of the research is to propose a computationally efficient and accurate pattern recognition algorithm for HCI. Deep learning attempts to model high-level abstractions (features) in data and build strong feature space for the recognition task. Neural network with five hidden layers was used and each layer can learn features at a different level of abstraction. However, training neural networks with multiple hidden layers was difficult in practice. At first, each hidden layer individually was trained in an unsupervised fashion using autoencoders. After training the first autoencoder, second autoencoder was trained in a similar way. The main difference is that features that were generated from the first autoencoder are used as the training data in the second autoencoder thus decreased the size of the hidden representation, so that the second autoencoder learns an even smaller representation of the input data. The original vectors in the training data had 101376 dimensions. After passing them through the first encoder, this was reduced to 10000 dimensions. After using the second encoder, this was reduced to 1000 dimensions. Likewise at the end, final layer was trained to classify 50 dimensional vectors into different image classes. The result for the deep neural network is improved by performing Backpropagation on the whole multilayer network. Finally, we observed that average test classification error for traditional neural network with supervised learning algorithm is 3.6% while the error for pre-trained deep neural network is 1.4%. We can conclude that unsupervised pre-training adds robustness to a deep architecture and it proposes computationally efficient and accurate pattern recognition algorithms for HCI.
Web Content Bookmarking Tool with Automatically Generated Titles
(Book of Abstracts, Annual Research Symposium 2014, 2014) Kumarika, B.M.T.; Dias, N.G.J.
With the development of technology, it has become essential to use the World Wide Web (WWW) for many day to day life activities. Whenever we navigate in the web, we are used to bookmark the most important and useful web pages so that those pages can be accessed later easily. But it is commonly noted that depending on the interests and relevant subject fields of a particular web user, the most useful focus can be on a specific part of a web page rather than the whole page. Therefore, the intention of this research is to introduce an efficient and novel method to bookmark all the specific contents or small part of text in each web page stored in one location corresponding to the interests of every individual web user. The most outstanding feature is that each bookmarked content will be well organized by the system suggested meaningful titles. To accomplish this task, key phrase extraction method which incorporates computational intelligence and web technologies is employed.