• V. Ajay Kumar Reddy
• V. Tejaswini
• M. Vamshi Krishna
• V. Nikitha

This paper proposes a robust framework for real-time accident de-tection under degraded CCTV conditions. The system integrates YOLO-based object detection with a CNN-based anomaly detection model to identify collisions, abnormal movements, and sudden changes in vehicle behavior. To ensure reliable performance in challenging environments such as low light, noise, rain, fog, motion blur, and low-resolution footage, a comprehensive preprocessing pipeline is applied, including brightness adjustment, noise reduc-tion, contrast enhancement, and frame stabilization. Furthermore, temporal smoothing and multi-frame validation techniques are incorporated to minimize false detections and im-prove prediction consistency. The proposed approach effectively combines spatial and temporal features to enhance detection accuracy and reliability. The system operates without requiring continuous internet connectivity, making it suitable for deployment in smart city surveillance systems, resource-constrained environments, and real-time emergency response applications.

Accident Detection, Poor Quality CCTV Footage, Anomaly Detection, CNN–YOLO Framework, Low-Light Enhancement, Real-Time Surveillance, Traffic Incident Prediction, Adverse Weather Vision, Video Pre-processing, Object Detection.

"ANOMALY-BASED ACCIDENT DETECTION AND PREDICTION UNDER DEGRADED CAMERA CONDITIONS USING CNN–YOLO FRAMEWORK", International Journal of Emerging Technologies and Innovative Research (www.jetir.org), ISSN:2349-5162, Vol.13, Issue 3, page no.h572-h578, March-2026, Available :http://www.jetir.org/papers/JETIR2603773.pdf

"ANOMALY-BASED ACCIDENT DETECTION AND PREDICTION UNDER DEGRADED CAMERA CONDITIONS USING CNN–YOLO FRAMEWORK", International Journal of Emerging Technologies and Innovative Research (www.jetir.org | UGC and issn Approved), ISSN:2349-5162, Vol.13, Issue 3, page no. pph572-h578, March-2026, Available at : http://www.jetir.org/papers/JETIR2603773.pdf