Intelligent Recognition of Traffic Signs Based on Improved YOLO v3 Algorithm

In recent years, assisted driving and autonomous technology have been paid more attention to by the public. Road sign recognition is of great practical significance for realization auto-driving technology. actual traffic environment, signs problems small detectable volume, low resolution, unclear characteristics, easy be disturbed environment. order better realize road recognition, this paper improves optimizes YOLO v3 network derived from structure algorithm, enhances data using color enhancement other technologies, original FPN algorithm 52 × 52. Then, secondary sampling output characteristic diagram 108 in used solutions solve these difficulties picture size image distortion. Use 5, 9, 13 fixed-size pools front surface control architecture, then characteristics are associated with so that inputs different sizes can obtain same output. Finally, we use intermediate class K group TT100K landmark set, reconsider parameters, compare set target determination such as model improved model. The results show compared traditional optimized has a significant improvement accuracy, speed, learning cost. When change FPS very small, recall rate accuracy will greatly improved. At time, detection algorithms, accurate faster accuracy.