Abstract: This research introduces a comprehensive deep learning-based system for restoring blurred images and videos while simultaneously enabling seatbelt detection from images for traffic safety monitoring. The proposed framework employs a Super-Resolution Network (SRN) for effective image deblurring and a Deep Blind Network (DBN) for handling motion blur in video sequences. Once the visual clarity is restored, The system includes a user-friendly interface where seatbelt detection can be triggered post-deblurring through a dedicated action button, making it suitable for applications in surveillance environments. Experimental evaluations show that the integration of deblurring and object detection significantly enhances recognition accuracy compared to processing blurred content directly. This unified approach not only recovers valuable visual information but also facilitates reliable enforcement of road safety regulations using restored visual evidence.

| DOI: 10.17148/IARJSET.2025.125220