Performance Analysis of Machine Learning Classification Algorithms in Static Object Detection for Video Surveillance Applications

Video surveillance system plays a pivotal role in automatic detection of abandoned luggage/bag in public places which causes threats to the public. As, the terrorist attacks are increasing world-wide, the detection and prevention of such attack is necessary to safeguard the people in public places. In this, a novel framework for the detection and classification of static object is proposed. In the proposed work first the static objects are identified and then it is classified to check the detected object is bag or anything else. In this study, the performance of machine learning techniques like Support Vector Machine (SVM), Artificial Neural Network (ANN), K-Nearest Neighbour, and Random Forest methods are analyzed. The performance is tested in standard (PETS 2006, PETS 2007 and AVSS i-LIDS) and custom datasets. The SVM and ANN produce best results in terms of classification and accuracy. Applications of various machine learning algorithms could clearly assist for identification and prevention of terrorist attacks in public places.