Abstract: In opposition to crime and criminals. The police are becoming less willing to respond to crime scenes unless there is visible confirmation, either by manned patrols or by electronic images from the surveillance cameras. The current systems do not classify routine and abnormal events.The proposed work is used for a variety of reasons, including live tracking, monitoring, classifying weaponry, and surveillance. In this work, real time image processing techniques are used to extract live surveillance footage from monitoring and identifying unusual events.The proposed project contains three processing modules. The first processing module uses Convolutional Neural Networks (CNN) for object identification, the second processing module handles the classification of weapons, and the third processing module handles monitoring and alert functions. A circular area will be monitored by CCTV, which will operate and be managed automatically. Before being implemented in such an environment, shape detection algorithms and object detection algorithms have been tested for accuracy in detection and analysis of processing time. The results provide the best accuracy in matching weapon and object types with names and shapes in predefined databases like ALEXNET. The proposed work will significantly lower crime rates, increase security in some regions, and shorten the time it
takes to apprehend offenders.
Keywords: Convolution Neural Network (CNN), Video Surveillance, Voice Assistance, Weapon Detection, Faster
Region based Convolution Neural Network (RCNN),picture segmentation.
|
DOI:
10.17148/IJARCCE.2023.12214
