F. H. K. Zaman

RECOGNITION OF HUMAN GAIT IN OBLIQUE AND FRONTAL VIEWS USING KINECT

This study describes the recognition of human gait in the oblique and frontal views using novel gait features derived from the skeleton joints provided by Kinect. In D-joint, the skeleton joints were extracted directly from the Kinect, which generates the gait feature. On the other hand, H-joint distance is a feature of distance between the hip joint with other skeleton joints. Prior to the gait feature extraction, the skeleton joints provided by Kinect were pre-processed in order to standardize the size of the skeleton image as well as to detect the gait feature within a full gait cycle. To classify gait patterns according to its own group, a multi-layer perceptron was employed in the pattern recognition stage. Results show that a perfect recognition of human gait (100%) was attained for the frontal view using the feature of H-joint distance at the optimal multi-layer perceptron (20 hidden units).

Ensembles of large margin nearest neighbour with grouped lateral patch arrangement for face classification

The use of local facial features is frequently adopted in many Nearest Neighbours (NN) approaches in face classification. These collections features are then individually classified against labelled features resembling an ensembles of simpler learners to improve prediction. In this paper, a new variant of ensembles of NN is proposed for classification of local features, namely ensembles of Large Margin Nearest Neighbour (soft-LMNN) classifier. Likewise, we proposea way to arrange local feature called Grouped Lateral Patch (GLP) to overcome the limitations of Single Lateral Patch (SLP). Since the performance of any NN method varies depending on the type of distance metrics used, we investigate the performance of ensembles of NN classifiers when Euclidean, Cosine, Manhattan, Chebychev and Minkowski distance metrics are used. From various experiments conducted, we found that soft-LMNN variant delivers best classification performance when compared against other NN variants, while Cosine and Manhattan distance metric performs best when used with locally normalized Gabor feature vectors and pixel intensity respectively. Our results also demonstrate that in general, ensembles of NN performs face classification nearly 14% more accurate than Support Vector Machine.