In 2023, the Ministry of Road Transport and Highways recorded 168,491 fatalities, with 44% (74,897) linked to two-wheeler riders. Shockingly, approximately 55,000 deaths were due to lack of helmet usage. To address this issue, India mandated helmet use, but enforcement relies heavily on manual efforts, increasing labor intensity. Various research efforts have explored solutions to enforce helmet compliance, often revolving around video and CCTV footage analysis by motor vehicle departments, focusing on post-event detection and penalization. In response, this study introduces a preventive solution: integrating a Deep Learning model (YOLO v8) with a dashboard camera to monitor helmet use in real-time. An IoT device like a Raspberry Pi connects the model to the vehicle’s ignition circuitry. The system allows ignition only if the rider wears a helmet. Continuous monitoring alerts the rider if the helmet is removed, shutting down ignition if ignored. The YOLO v8 model was trained on a newly created helmet dataset which consists of 400 close-up images of riders with helmet and 400 images without helmet which were captured in such a way that it resembles the images which would be captured from a handle bar dash-cam. The existing helmet detection datasets had helmet images captured from CCTV footages which couldn’t be used for this use-case. Also it consisted of industrial helmets and bicycle helmets which couldn’t be considered as bike helmets. The proposed system, trained on a specialized helmet dataset, achieves a mean Average Precision (mAP) score of 0.83 and Intersection over Union (IoU) score of 0.79 with an processing speed of 30 fps. This comprehensive approach fosters a culture of real-time helmet compliance, offering a promising strategy to reduce two-wheeler fatalities and promote safer roads.