Key points are not available for this paper at this time.
In the contemporary landscape, two-wheelershave emerged as the predominant mode oftransportation, despite their inherent risk due to limitedprotection. Disturbing data from 2020 reveals a dailytoll of 304 lives lost in India in road accidents involvingtwo-wheeler riders without helmets, emphasizing theurgent need for safety measures. Recognizing thecrucial role of helmets in mitigating risks, governmentshave made riding without one a punishable offense,employing manual strategies for enforcement withlimitations in speed and weather conditions. In today'sworld of advancing technology, we can leveragethe power of computer vision and deep learning totackle this problem. This can eliminate the need forconstant human surveillance to be kept on riders andcan automate this process, thus enforcing law andorder as well as making this process efficient. Ourproposed solution utilizes video surveillance and theYOLOv8 deep learning model for automatic helmetdetection. The system employs pure machine learningto identify helmet types with minimal computation costby utilizing various image processing algorithms. Oncethe helmet-less person is detected, the number platecorresponding to the rider's motorcycle is also detectedand extracted using computer vision techniques. Thisnumber plate is then stored in a database thus allowingfurther intervention to be done in this matter by theauthorities to ensure penalties and enforce safety rulesproperly. The model developed achieves an overallaccuracy score of 93.6% on the testing data, thusshowcasing good results on diverse datasets.
Prakash-Borah et al. (Wed,) studied this question.