HOG Based Surveillance System for Chain Snatching Detection

In certain countries, the usage of closed-circuit television (CCTV) cameras for public safety and crime prevention has increased recently. Traditional CCTV systems, on the other hand, have a reputation for being ineffective at spotting suspicious activity. By utilizing artificial intelligence and machine learning techniques to identify odd behavior, the Abnormal Activity Detection work seeks to get over this constraint. A sophisticated methodology is used in the study to identify suspicious activity in CCTV footage. The proposed system employs two methods for abnormal activity detection in chain snatching incidents. These methods utilize the HOG and SIFT features and are implemented using various classifiers, including random forest, XGBoost, logistic regression, decision tree, and SVM. The system is designed to analyze video frames and detect chain snatching incidents in real-time. The proposed system is novel in its ability to be trained on various types of video footage with different categories, making it a versatile tool for crime prevention and public safety. Ultimately, the Abnormal Activity detection system offers a comprehensive approach to CCTV surveillance that enhances public safety and aids in crime prevention.