Multi-mode Summarization of Surveillance Videos using Supervised Learning techniques

Video summarization has been a prevailing area due to the wide use of surveillance systems. Extracting the contents from these video frames for domain-specific usage has become a tedious task. There are many approaches for the augmentation and detection of video content. A single platform for generating a summary of the surveillance videos will be revolutionary as security surveillance is becoming challenging day by day. This work is based on the various literature available for managing the video content especially for extracting the face, activity, and detecting the Object of Interest (OoI). The agents are identified based on an extensive literature survey on various approaches for object detection, activity recognition, and identity verification. The survey has helped to find out the best-performing algorithm in each of the above-mentioned domain. The work has proposed a unique architecture, Multi-mode Summarization of Survellance video System(MSSVS) where the observer can select agents, and based on this the metatag of the activity can be generated.