Adaptive algorithm to identify anomalies in moving objects using computer vision

In this document, an adaptive algorithm is proposed to identify anomalies in moving objects, such as pedestrians, cars, motorcyclists and cyclists. The anomalies detected by this algorithm are: occlusions, an object partially entering or exiting a frame, alterations in the object velocities or collisions between two or more objects in the frame. The classification between frames with anomaly and no anomaly is achieved by finding an adaptive threshold that depends on the video sequences.