Systematic computational analysis of data based on video-object tracking for surveillance

• Published in: 2017 2nd IEEE International Conference on Recent Trends in Electronics, Information & Communication Technology (RTEICT) pp. 2133 - 2138
• Main Authors: Bhat, Nagaraj, Eranna, U., Mahendra, B. M., Sonali, Savita
• Format: Conference Proceeding
• Language: English
• Published: IEEE 01-05-2017
• Subjects: AOD; Benchmark testing; i-LiDS AVSS; Security; Streaming media; Surveillance
• DOI: 10.1109/RTEICT.2017.8256977

Abandoned objects in public places is one of the typical surveillance breaches. Detecting an abandoned object in surveillance video is very important to forecast terrorist activity. This work aims to develop a modular system with several individual stages where in at each stage different algorithm is employed. The overall task is to detect abandoned object in a video stream. This has been implemented in Math Work's MATLAB integrated development environment. The performance of the system is evaluated on test videos from standard publically available datasets and also custom dataset. The Abandoned Object Detection system is tested for two different datasets-publically available i-LiDS AVSS and custom dataset. The metric called system performance used to evaluate our system provided 85.71% result for AVSS dataset and 75% for custom dataset, with overall system performance reaching up to 78.125%.