People recognition in videos

Abstract

The approach developed intends to automate the process of video surveillance systems which are a part of many industries such as security sensitive areas, shopping malls, highways, smart homes and offices. It involves interdisciplinary research such as artificial intelligence and image Processing. An image extracted from a video is dependent on certain attributes that affect its quality such as the light, angle, speed of the video being selected and the quality of camera. The system uses a video to be processed and then further follows the detection of face from the particular frame through Support Vector Machine (SVM) since the dataset has been trained with HOG. HOG showed better system performance as compared to CNN, so this system uses HOG. i