Enhanced Surveillance Through YOLOv3-Based on Deep Learning for Real-Time Weapon Detection
A Shalini, Satya Charan V, Khyrul Mateen Khan, K. Nithin, Vaishnavi Shantagiri, B Rohan Srivatsav · 2025
For public safety requirements, the use of machine learning in developing weapon detection in surveillance footage is essential. We apply YOLO v3, which is based on convolutional neural networks (CNN) for image and video analysis that helps in distinguishing weapons from non-weapon objects. We train our models on a large variety of datasets to increase their accuracy, and we utilize transfer learning strategies to lower the number of false positives and negatives. According to the early findings, machine learning holds great promise for improving real-time weapon detection. This work presents a low- cost system design in case study form, demonstrating applicability to large security projects for diverse environments like airports or public events where different types of threats have to be addressed with limited resource allocation but still effective enough towards ensuring safety and security within those areas.