Real-Time Crime Detection and Tracking on Edge Devices Using RCNN - BiGRU Models

Criminal problems are large, unpredictable, and difficult to solve. Delayed detection, identification, and response to criminal events are substantial constraints. Many feel that police forces (PFs) use internal communication strategies that don't engage third parties in their investigations, causing this delay. A three-step strategy comprising training, feature extraction, and model face tracking addresses these issues. Face-tracking employs Optical Flow, which uses picture motion patterns, to detect moving objects. Feature extraction—building meaningful variable combinations to efficiently explain data—is crucial to the process. The Levy Flight Hybrid Recurrent Neural Network (RNN) trains models using its new approach to complex datasets. The proposed model improves accuracy by 93.24%. Region Proposal Networks (RPNs) and standard RNNs can't match this performance. Optical Flow and Levy Flight Hybrid RNN cover gaps in criminal event identification and intervention. The results suggest that third-party data and enhanced feature extraction can improve crime detection and response.