A Novel Approach for Acrimonious Attack and Violence Detection Using AI and Edge Devices

Abstract- -With the rapid growth and development of the application of artificial intelligence systems. Acrimonious attacks and violence detection are hot topics of this application nowadays. An acrimonious attack and violence detection are related to a criminal attack or something happening between two people or a criminal activity. In traditional methods, humans detect violence and attacks using a surveillance system, but methods and processes are changing with the growing development phase in intelligence systems. In this paper, a novel approach has been proposed, Image to Image stable diffusion for creating the synthetic data and pose estimation and object detecting approach is detecting the attacks on the victims, where one stage YOLO V5 object detector is being used for violent object detection, objects like baseball bats, knives, and pistols and MediaPipe for action detection. Further, LSTM classifies the action attacks with violent objects. Now Edge device ensemble with a dash camera and the whole proposed model are deployed in the device. Moreover, this study holds potential for ensuring precise detection of violent incidents. It introduces a robust system that attains remarkable accuracy, with an overall model precision of 94.0%. The results showcase the model’s improved ability to precisely detect violent objects, especially in effectively identifying violence through the implemented detection algorithm.

DOI: /10.61463/ijset.vol.12.issue1.111