Automated Crowd Abnormality Detection and Segmentation Using Machine Learning Techniques
Spoorthi S K, Prasanna Kumar M J · 2025
Crowd analysis has evolved as a highly prominent research subject within the field of computer vision. For public safety, automated analysis of crowd dynamics from surveillance video is essential since it allows for the identification of potentially dangerous crowd formations and the tracking of their movements. We are all aware of the various difficulties that come with big parties. Terrorist attacks pose a serious threat to our nation since they frequently position bombs in busy places, which causes large-scale casualties. Furthermore, thieves usually operate in crowded areas, using the confusion to their advantage to conduct crimes undetected. Effective crowd analysis is crucial given these hazards. This work presents a deep learning framework for crowd behaviour monitoring and control, with the objective of averting violent episodes and other detrimental behaviour’s that frequently occur in such environments. The suggested method aims to detect unusual crowd behaviour by utilizing cutting-edge deep learning algorithms. This proactive approach improves public safety and reduces possible hazards to society.