Video content analysis using convolutional neural networks
Inad A. Aljarrah, Duaa Mohammad · 2018
Video content analysis has been an active research area due to the huge number of real-life applications that utilize it in a way or another to perform their functionalities. Reviewing videos recorded by video surveillance systems is an area where video content analysis can be handy. Instead of rewinding over hours of recorded video to spot an action, an automated content analysis system that produces a searchable text file that summarizes the video content is proposed. In this work, video surveillance content is analyzed using object classification. Objects appearing in the video are detected and classified using a convolutional neural network model. A text file that contains the classes of detected objects and the time of appearance is generated for later search. To speed up this computational heavy process, only (I) frames are processed.