Research on obstacle detection algorithm based on YOLOX
Jianshu Gao, Pu-Ning Zhang · 2022 7th International Conference on Intelligent Computing and Signal Processing (ICSP) · 2022
Aiming at the problem that the computing power of current computing equipment cannot meet the needs of emerging algorithms, an improved YOLOX obstacle detection method is proposed. This algorithm adds an attention mechanism to the YOLOX detection algorithm based on the original algorithm, which reduces the computing scale of hardware devices and achieves high speed and high accuracy for obstacle recognition. By improving the feature extraction method, the detection effect of the target in the video is improved. The experimental results show that the detection accuracy of the improved algorithm is increased from the original 89.47% to 91.21%, and the mAP (mean Average Precision) index is increased from the original 83.4% to 84.1%. The improved result is better than the original algorithm.