Violence Detection System Using Resnet

In recent years, human motion analysis has become the center of video processing, especially in motion detection and security surveillance. When performing specific tasks, it is always a tool to minimize human resources. However, earlier methods of detection often descriptors are extracted around spatiotemporal attention points or statistical features of motion regions are extracted, resulting in limited ability to effectively detect violence in videos. This paper is intended to use a Pre-trained Convolutional Neural Network Model (ResNet), where the video stream is sent to the ResNet together with a single shot detector (S S D) to detect violence and objects in the video. In addition, if an intense incident is identified, the relevant departments will be notified. This proposed methodology would be beneficial in monitoring scenarios such as detention center & asylum. Also, on mobile phones with cameras.