YuHengsss / YOLOV

This repo is an implementation of PyTorch version YOLOV Series
Apache License 2.0
328 stars 47 forks source link

请问下如何改进静态目标检测器使其更适合于视频目标检测? #95

Open xingguang12 opened 3 months ago

xingguang12 commented 3 months ago

静态目标检测器(YOLO系列)型无法利用视频流中时间信息及上下文信息,导致在视频场景下的检测效果不太好(视频数据存在的画面模糊、物体遮挡、镜头高速移动、物体罕见姿势等),因为算力受限的原因无法使用视频目标检测模型(一张显卡需要处理多路视频流),现在我想改进YOLOv8s模型添加一些后处理模块来获取视频数据的时间信息和上下文信息,但是没有啥好的思路,请问一下您有一些建议或者推荐一些相应的论文(或修改代码)吗?

YuHengsss commented 3 months ago

You may refer to https://github.com/AlbertoSabater/Robust-and-efficient-post-processing-for-video-object-detection. But it's for the offline video object detection. Classic post-processing methods for video object detection such as Seq-NMS may help you.