Query- Focused Content Retrieval from Surveillance videos: Techniques and Challenges

The rocketing demand of security and safety has made video surveillance to be an indispensable part of the research. Over few decades a remarkable technological growth has been observed in the area of security and safety. A huge number of surveillance cameras are mounted everywhere such as hospitals and financial institutions. The growth of artificial intelligence and machine Learning is giving a digital brain and the capability to analyze the videos captured by those cameras. Contents of the video can be retrieved using some queries and the queries can be an object, a text, an example, a shot or a clip. The content based retrieval is mainly categorized into two. Intelligent content Based retrieval and Non Intelligent content Based retrieval. The Intelligent video retrieval uses a neural network that studies the pattern in the videos and the classify them accordingly. Non Intelligent systems uses some primitive methods such as text, clip etc to retrieve the contents from the videos. This paper analyses various techniques to retrieve the contents from the videos. Convolutional Neural Network, Genetic Algorithm, Pattern Recognition,YOLO-3,Darknet