search

positive666 / yolo_research

based on yolo-high-level project (detect\pose\classify\segment\):include yolov5\yolov7\yolov8\ core ,improvement research ,SwintransformV2 and Attention Series. training skills, business customization, engineering deployment C
GNU General Public License v3.0
756 stars 146 forks source link

为什么采用yolov5-p6.yaml的修改版本训练不起来？（mAP50才0.00095) #89

Closed xingguang12 closed 1 year ago

xingguang12 commented 1 year ago

❔Question

Snipaste_2023-03-01_10-10-01 采用yolov5-p6.yaml的修改版本，只是将depth_multiple改成0.33，width_multiple改成0.3，就训练不起来 Snipaste_2023-03-01_09-57-15 网络架构图 Snipaste_2023-03-01_10-14-17

Additional context

positive666 commented 1 year ago

你基于官方YOLO5修改的结构吗

xingguang12 commented 1 year ago

你基于官方YOLO5修改的结构吗

是的啊， https://github.com/ultralytics/yolov5/blob/master/models/hub/yolov5-p6.yaml 所进行的改动：把yolov5-p6的depth_multiple改成0.33，width_multiple改成0.3 改动原因：复现所看的论文 论文链接：https://www.mdpi.com/2076-3417/13/3/1746

positive666 commented 1 year ago

你基于官方YOLO5修改的结构吗

是的啊， https://github.com/ultralytics/yolov5/blob/master/models/hub/yolov5-p6.yaml 所进行的改动：把yolov5-p6的depth_multiple改成0.33，width_multiple改成0.3 改动原因：复现所看的论文 论文链接：https://www.mdpi.com/2076-3417/13/3/1746

（不知道你是否排除了一些明显的问题，比如你换了原版模型是正常的精度） 最近分身乏术了 你描述的这个是你把模型压缩了 超参数学习率调整 ，我MARK下 不过要等我把V8整理好了再说

github-actions[bot] commented 1 year ago

This issue has been automatically marked as stale because it has not had recent activity. It will be closed if no further activity occurs. Thank you for your contributions.