PaddleDetection icon indicating copy to clipboard operation
PaddleDetection copied to clipboard

混淆矩阵中错误的分类的缺陷如何确定是哪些图片贡献的特征

Open xillee opened this issue 1 year ago • 3 comments

问题确认 Search before asking

  • [X] 我已经搜索过问题,但是没有找到解答。I have searched the question and found no related answer.

请提出你的问题 Please ask your question

我有6个检出类别,其中有两个类别比较近似,相互之间有误检,且置信度比较低,因此我想确认图片的缺陷的检出置信度是由哪些图片贡献的,为什么会置信度比较低。

xillee avatar Oct 12 '24 06:10 xillee

您好, 可以保存结果再过滤下低分框。--save_results=True

Sunting78 avatar Oct 14 '24 11:10 Sunting78

请问保存的低分框怎么用?低分框即为误检图片的贡献框吗?

xillee avatar Oct 15 '24 09:10 xillee

不一定 可以把预测结果框和这张图的GT在进行同类iou的匹配 如果没有匹配到就是误检

Sunting78 avatar Oct 16 '24 12:10 Sunting78

The issue has no response for a long time and will be closed. You can reopen or new another issue if are still confused.


From Bot

TingquanGao avatar Nov 17 '24 03:11 TingquanGao