AI-Based Smart Traffic Police Constable

Road safety enforcement depends on officers observing large volumes of traffic, which is slow, inconsistent and difficult to extend to every junction. This work presented a prototype AI-based smart traffic constable that helps with one part of the task: automatically screening images and live camera frames for riders without helmets and for triple riding, and drawing an officer's attention to them. The implemented system is a Python and Tkinter dashboard built around a fine-tuned nano-scale Ultralytics YOLO detector with nine classes. OpenCV supplies camera access and image output. The dashboard displays annotated detections, per-class counters, a mean confidence value and a SAFE or VIOLATION status, saves annotated images, and sends an e- e-mail alert with an image attachment under a cooldown. The contribution of the project is a complete, working end-to-end baseline, from model to interface to notification, that is easy to run on ordinary hardware and easy to extend. Its limitations are equally clear. The project supplies no quantitative evaluation, so no claim of detection accuracy, accident reduction or enforcement benefit can be made. Violation detection is a label-presence rule, with no tracking, plate recognition, location or evidence database. The class named triple does not appear to match the rule that expects triple seat. Live-mode inference runs in the interface thread, every live frame is written to disk, and the e-mail configuration is hard-coded. In real deployment, further considerations include camera placement, night-time performance, dataset representativeness, data protection for captured images, legal admissibility of automated evidence, and regular re-validation of the model.

[1]         J. Redmon, S. Divvala, R. Girshick, and A. Farhadi, "You Only Look Once: Unified, realtime object detection," in Proc. IEEE Conf. Computer Vision and Pattern Recognition (CVPR), Las Vegas, NV, USA, 2016, pp. 779-788, doi: 10.1109/CVPR.2016.91. [2]         Ultralytics, "Ultralytics YOLO documentation and source code." [Online]. Available: https://docs.ultralytics.com and https://github.com/ultralytics/ultralytics (version-specific details for the YOLO26 model family to be verified against the documentation before submission). [3]           G. Bradski, "The OpenCV Library," Dr. Dobb's Journal of Software Tools, 2000. [4]           Python Software Foundation, "tkinter - Python interface to Tcl/Tk," Python 3 Documentation. [Online]. Available: https://docs.python.org/3/library/tkinter.html [5]           A. Neubeck and L. Van Gool, "Efficient non-maximum suppression," in Proc. 18th Int. Conf. Pattern Recognition (ICPR), Hong Kong, 2006, vol. 3, pp. 850-855, doi: 10.1109/ICPR.2006.479. [6]         T.-Y. Lin et al., "Microsoft COCO: Common objects in context," in Computer Vision - ECCV 2014, Lecture Notes in Computer Science, vol. 8693, Springer, 2014, pp. 740-755, doi: 10.1007/978-3-319-10602-1_48. [7]           A. Bewley, Z. Ge, L. Ott, F. Ramos, and B. Upcroft, "Simple online and real time tracking," in Proc. IEEE Int. Conf. Image Processing (ICIP), Phoenix, AZ, USA, 2016, pp. 3464-3468, doi: 10.1109/ICIP.2016.7533003. [8]         N. Wojke, A. Bewley, and D. Paulus, "Simple online and real time tracking with a deep association metric," in Proc. IEEE Int. Conf. Image Processing (ICIP), Beijing, China, 2017, pp. 3645-3649, doi: 10.1109/ICIP.2017.8296962. [9]           R. Smith, "An overview of the Tesseract OCR engine," in Proc. 9th Int. Conf. Document Analysis and Recognition (ICDAR), Curitiba, Brazil, 2007, pp. 629-633, doi: 10.1109/ICDAR.2007.4376991. [10]           M. Everingham, L. Van Gool, C. K. I. Williams, J. Winn, and A. Zisserman, "The PASCAL Visual Object Classes (VOC) challenge," Int. J. Computer Vision, vol. 88, no. 2, pp. 303-338, 2010, doi: 10.1007/s11263- 009-0275-4.