Motorized two-wheelers are a prevalent and economical means of transportation, particularly in the Asia-Pacific region. However, hazardous driving practices such as triple riding and non-compliance with helmet regulations contribute significantly to accident rates. Addressing these violations through automated enforcement mechanisms can enhance traffic safety.
In this paper, we propose DashCop, an end-to-end system for automated E-ticket generation. The system processes vehicle-mounted dashcam videos to detect two-wheeler traffic violations. Our contributions include: (1) a novel Segmentation and Cross-Association (SAC) module to accurately associate riders with their motorcycles, (2) a robust cross-association-based tracking algorithm optimized for the simultaneous presence of riders and motorcycles, and (3) the RideSafe-400 dataset, a comprehensive annotated dashcam video dataset for triple riding and helmet rule violations.
Our system demonstrates significant improvements in violation detection, validated through extensive evaluations on the RideSafe-400 dataset.
Our SAC module improves rider-motorcycle association by learning to detect cross-object class, outperforming IOU-based methods. The joint tracking approach reduces false positives in crowded scenes.
Frame-level detection approach achieves superior F1-scores in helmet violation detection, demonstrating better balance across different ROI extraction methods compared to existing solutions.
Our SAC module achieves superior association accuracy compared to geometric heuristic methods by learning directly in image space rather than relying solely on bounding box coordinates.
Joint tracking of rider-motorcycle instances demonstrates better performance across HOTA, MOTA, and IDF1 metrics compared to independent tracking and post-hoc aggregation approaches.
For a given R-M track, the labelling (TP, FP, FN) at various stages of system is used to determine the final E-ticket level label of the track. E.g., a True Positive (TP) prediction at all stages is considered a TP for E-ticket generation system.
The automated system achieves an F1-score of 72.18%, reflecting good overall performance. In practice, traffic enforcement personnel review E-Tickets and corresponding evidence before issuance. This human-in-the-loop approach eliminates false positives, raising the F1-score to 82.05%, improving system reliability.
The RideSafe-400 dataset captures a wide range of real-world scenarios that challenge traffic violation detection systems. These include crowded scenes with multiple motorcycles, varying lighting conditions, partial occlusions, and diverse viewing angles. The dataset features comprehensive annotations for triple riding violations and helmet rule compliance across a range of urban and suburban environments.
We provide both frame-level and track-level annotations for rider-motorcycle pairs, triple riding instances, and helmet violations. The dataset includes detailed license plate annotations with format information, enabling comprehensive evaluation of our automated E-ticket generation system. These annotations support thorough evaluation of detection, tracking, and violation recognition tasks.
Coming Soon!
@InProceedings{Rawat_2025_WACV,
author = {Rawat, Deepti and Gupta, Keshav and Roy, Aryamaan Basu and Sarvadevabhatla, Ravi Kiran},
title = {DashCop: Automated E-Ticket Generation for Two-Wheeler Traffic Violations using Dashcam Videos},
booktitle = {Proceedings of the Winter Conference on Applications of Computer Vision (WACV)},
month = {February},
year = {2025},
pages = {5387-5397}
}
deepti.rawat@research.iiit.ac.in