An investigation of videos for abnormal behavior detection

• DOI: 10.1016/j.procs.2023.01.202
• Publication Date: 2023-01-30T19:57:13Z
• Authors: Akshita Patwal; Manoj Diwakar; Vikas Tripathi; Prabhishek Singh
• Keywords: Identification; Extractor; Feature (linguistics)

The identification of abnormal behavior has several applications. There are ways, ranging from classical to deep learning based. It may be used monitor campuses, banks, transportation, and airports. In many circumstances, the context determines whether real-life events common or unusual. Recent video surveillance anomaly detection systems good enough, but they come at a significant computational cost require particular hardware resources. When it comes real-time detection, extra emphasis must paid on lowering model complexity, which causes memory demands. This study attempts find real-world abnormalities in CCTV recordings, such as violence, detention, property destruction, assault, burglary, blast theft. impact these public safety is enormous. also provides low-cost algorithm for detecting crowd irregularities. proposed uses convolutional neural network based DenseNet121 feature extractor. Our suggested framework an AUC 86.63 percent UCF-Crime dataset obtains new stateof-the-art performance