混合现实的目标识别技术和YOLO算法的对比

Comparison Between Target Recognition Technology of Mixed Reality and YOLO Algorithm

针对混合现实技术中存在的识别距离过短、识别时间过长等问题,利用深度学习中的YOLO算法在目标检测上有着很快处理速度和很长识别距离的优点,尝试把YOLO算法的优势结合到混合现实技术中,以提高混合现实的工作效率,提出了一种技术方案。通过对混合现实的特征点匹配技术和深度学习的YOLO算法的比较,借助深度学习YOLO算法快速识别和远距离识别的优点来改进混合现实的识别技术,从而使混合现实技术达到减少特征点匹配的时间、增加识别距离、更容易识别三维物体等目的。

Aiming at the problems of too short recognition distance and too long recognition time in mixed reality technology, taking advantage of the fast processing speed and long recognition distance of YOLO algorithm in depth learning in target detection, this paper attempts to combine the advantages of YOLO algorithm into mixed reality technology to improve the work efficiency of mixed reality, and proposes a technical scheme. By comparing the feature point matching technology of mixed reality with the YOLO algorithm of depth learning, the recognition technology of mixed reality is improved by virtue of the advantages of fast recognition and long-distance recognition of the YOLO algorithm of depth learning, so that the mixed reality technology achieves the purposes of reducing the matching time of feature points, increasing the recognition distance, and making it easier to recognize three-dimensional objects, etc.