High-Accuracy Object Detection Using Multi-view Video at Road Intersections
High-Accuracy Object Detection Using Multi-view Video at Road Intersections
复制标题
DOI：
10.1109/itc-cscc55581.2022.9894885
复制
发表时间：
2022-07
期刊：
影响因子：
--
通讯作者：
Urumu Ihara;Takafumi Katayama;Tian Song;T. Shimamoto
中科院分区：
文献类型：
--
作者：
Urumu Ihara;Takafumi Katayama;Tian Song;T. Shimamoto
在本文中，我们提出了通过多视频视频补充彼此信息的算法。视频在同一位置的像素位置相应相关。分别使用Yolo和DeepSort推断两个观点视频，并且彼此的结果相互使用。通过提出的方法，我们将未发现的率降低了97％以上。
In this paper, we proposed algorithms that complement each other's information by multi-view video. The pixel positions of the video at the same location are correspondingly related. The two viewpoint videos are inferred using YOLO and DeepSORT, respectively, and the results of each are used against each other. We reduced the undetected rate by more than 97% with the proposed method.