1Mrs. K. DURGA BHAVANI, 2K. VARSHA, 3G. RUTHVIK REDDY, 4A. SUVIKRAM 03 05 Helmet Track AI is an advanced deep learningbasedsystem designed to enhance road safety byautomatically detecting helmet usage amongmotorcyclists in realtime traffic environments Therapid increase in twowheeler usage hassignificantly contributed to road accidentsparticularly due to noncompliance with helmetregulations To address this issue the proposedsystem employs Convolutional Neural NetworksCNNs integrated with MultiTask LearningMTL to simultaneously perform motorcyclistdetection and helmet classification Unliketraditional systems that process detection andclassification separately this unified approachreduces computational redundancy and improvesinference speed The system is trained on diverserealworld datasets enabling it to performeffectively under varying lighting conditionsweather scenarios camera angles and motion blurIt utilizes realtime video streams captured throughsurveillance cameras and processes them usingadvanced computer vision techniques for accuratedetection The integration of YOLObased objectdetection further enhances performance byenabling fast and efficient localization of riders andhelmets Experimental results indicate thatHelmetTrack AI achieves high precision recall andaccuracy compared to conventional modelsAdditionally the system supports edge deploymentmaking it suitable for smart city applications andautomated traffic law enforcement By providingrealtime monitoring violation detection and datastorage capabilities the system contributes toimproved compliance and reduced accident ratesOverall HelmetTrack AI offers a scalable efficientand intelligent solution for modern trafficsurveillance systems Copyright (c) Journal of Science Engineering Technology and Management Science. All rights reserved 2026