Gesture Recognition using template matching, motion history image and machine learning. The project is basically divided into 3 phases involving segmentation, data collection and learning. Segmentation is done using fast and robust template matching. Details of template matching are mentioned in the proposal. Phase 2 will involve deciding various gestures and corresponding data collection of these gestures from the mentioned NewsScape database. Phase 2 will also involve feature extraction. Three different kinds of features will be extracted. Firstly the template features used to segment the hand and head will be used , secondly the features generated using motion history image of the segmented hand will be used and finally the relative positions of the hands and heads will be used. These 3 types of features will be stored and used for learning. Learning will involve using 2dSvd for dimensionality reduction without losing time information. Various classifiers will b used to learn the reduced data and the one the the highest accuracy will be chosen.