Parallel Deep Learning Framework for Video Surveillance System

In today’s world, the security of every individual has become an important aspect. There is a need for constant monitoring in public places. A Manual operating camera system unreliable and very basic poor method this purpose. Intelligent Video Surveillance approach where multiple CCTVs constantly record scenes proper algorithms are deployed order to detect monitor activities. Deep Learning frameworks like Kera’s, YOLO, Convolutional Neural Networks or backbones image detection VGG16, Mobile net, Resnet101 have been used human weapon detection. The paper focuses on deep learning techniques threading collectively develop Parallel Framework that aims at striking right balance between accuracy performance stability. Threading terms implementation uniquely proposed Dynamic Selection Algorithm uses two object switches them based queue status achieving designed logistic regression filter also implemented boosts performance.