Anomaly detection based on superpixels in videos

The article introduces a novel method for anomaly detection in videos using superpixels, addressing the limitations of traditional grid-based methods. It defines anomalies as rare events and highlights the importance of detecting such events in various applications like intelligent surveillance and behavior analysis. The method divides video frames into superpixels and uses motion histograms to describe object motion, improving the accuracy and efficiency of anomaly detection. An online scene model is employed to select superpixels in motion regions, further enhancing detection performance. The article also discusses the advantages of combining low-level grid histograms and mid-level superpixel histograms for anomaly detection. Experiments on public datasets demonstrate the effectiveness of the proposed method, making it a valuable contribution to the field of computer vision and anomaly detection.