An Automated Approach to Detect and Label Abandoned Objects from Videos using Generalized ROIs

In this paper, abandoned object detection has been performed and represented through named labels using generalized ROIs. In the surveillance system, various activities are monitored for security purposes, among which Abandoned Object Detection is one. The proposed method is performed on various existing benchmark datasets such as CAVIAR, ABODA, and TCD for the verification and validation of work. Initially, the model is trained using the generalized ROIs labels to create ground truth data. The eigenvalue distance method is used for identifying the change in Point-Tracker with generalized ROI to generate the results. The experimental results show that the generalized ROI gives 50-95% improvement in terms of accuracy in the detection of Abandoned Object in the various samples of the Datasets as compared to the existing techniques.