System for automatic detection and classification of cars in traffic

Objective : To develop a system for automatic detection and classification of cars in traffic the form device autonomic, real-time car detection, license plate recognition, color, model, make identification from video. Methods: Cars were detected using You Only Look Once (YOLO) v4 detector. The YOLO output was then used next step. Colors classified k-Nearest Neighbors (kNN) algorithm, whereas models makes identified with single-shot detector (SSD). Finally, plates OpenCV library Tesseract-based optical character recognition. For sake simplicity speed, subsystems run on an embedded Raspberry Pi computer. Results: A camera mounted inside windshield to monitor front camera. processed camera’s video feed provided information plate, make, model observed car. Knowing number provides access details about owner, roadworthiness, or place reports missing, as well whether matches Car saved file displayed screen. tested images videos. accuracies (using 8 classes) 88.5% 78.5%, respectively. color full recognition 71.5% 51.5%, operated at 1 frame per second (1 fps). Conclusion: These results show that running standard machine learning algorithms low-cost hardware may enable traffic. However, there is significant room improvement, primarily Accordingly, potential improvements future development are proposed.