A Framework for the Video Surveillance Suspicious Activity Detection

Video surveillance is globally considered to be of considerable importance. Recent advances have resulted in notable improvements in the incorporation of artificial intelligence, machine learning, and deep learning techniques into video surveillance devices. The utilization of combinations and distinct frameworks facilitates the differentiation of various questionable behaviors through real-time image analysis. Human behavior is inherently unpredictable, making it difficult to determine whether it is suspicious or typical. This study characterized human actions into two categories: normal and suspicious. Normal actions include sitting, strolling, running, waving hands, etc., while arrest, abuse, shoplifting, etc., are examples of suspicious actions. This study used a convolutional neural network, achieving 97.96% accuracy on the CIFAR-100 dataset, demonstrating its effectiveness in recognizing and categorizing various activities, and paving the way for improved surveillance and security applications. Future work will focus on further refining the model and expanding its capabilities to include real-time video analysis, allowing more dynamic responses to potential threats and enabling faster decision-making in critical situations. Additionally, the integration of advanced algorithms for behavior prediction could further enhance the model's performance in complex environments.