A new study published in IEEE Access introduces an AI-powered surveillance system designed to enhance public safety by detecting high-risk individuals in real-time. With violent crimes, particularly those involving repeat offenders, posing a significant threat to vulnerable areas like schools and childcare centers, this system uses deep-learning facial recognition to identify criminals from surveillance footage. The system processes video frames at over 30 FPS using down-sampling and face tracking, ensuring real-time performance. It combines detection confidence and standard scores to minimize errors, achieving 90% accuracy and a 94.3% F-1 score. By detecting criminals early and notifying authorities, it helps prevent crimes before they occur.
The full paper is available under a Creative Commons license: https://ieeexplore.ieee.org/document/10143202/
Leave a Reply