Road Anomaly Detection

Please Cited paper. https://doi.org/10.1038/s41598-025-29718-4

The Road Anomaly Dataset (RAD) is a benchmark dataset developed to advance research in automated surveillance systems, intelligent transportation systems, and video-based anomaly detection. It specifically addresses the critical limitations of conventional CCTV monitoring, where vast amounts of continuously generated video data from public and private surveillance networks cannot be effectively analyzed by human operators in real time. This gap has created a strong demand for intelligent, automated systems capable of detecting, classifying, and responding to abnormal events with high accuracy and minimal delay.
The RAD dataset has been carefully constructed for experimental and academic research purposes. It comprises approximately 70 real-world video sequences along with thousands of extracted frames, providing a rich and diverse foundation for training and evaluating modern computer vision and deep learning models. The dataset incorporates videos recorded under multiple spatial and temporal configurations, including 1920 × 1080 at 25 fps, 848 × 480 at 30 fps, and 640 × 480 at 24 fps, ensuring variability in resolution, motion patterns, lighting conditions, and scene complexity. This diversity enhances the robustness and generalization capability of models trained on the dataset. RAD covers a set of significant road-related anomalous events that are critical for public safety and traffic monitoring applications. These include road accidents, vehicle fire incidents, physical altercations (fighting cases), snatching incidents involving armed threats (gunpoint situations), and road surface defects such as potholes. Each category reflects real-world challenges frequently encountered in urban traffic environments, making the dataset highly relevant for practical deployment scenarios. Overall, the RAD dataset serves as a comprehensive and valuable research resource for the development and benchmarking of advanced AI-driven surveillance solutions. It facilitates progress in real-time anomaly detection, video understanding, and smart city applications, contributing toward safer transportation systems, improved incident response mechanisms, and enhanced situational awareness in intelligent surveillance frameworks.