Skip to content
New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

目标检测预训练模型问题 #193

Closed
cctx1 opened this issue Apr 13, 2024 · 2 comments
Closed

目标检测预训练模型问题 #193

cctx1 opened this issue Apr 13, 2024 · 2 comments
Assignees
Labels
question Further information is requested

Comments

@cctx1
Copy link

cctx1 commented Apr 13, 2024

在使用目标检测中的PPYOLOv2模型时,报warning:”预训练的权重参数与实际模型的参数形状不匹配“,如图1。貌似是因为这个导致loss震荡,最后导致训练结果不理想。求助解决这个warning的方法。图2为训练完整代码。使用的数据集为DIOR数据集,共分了20类,标签为VOC格式
图1
P0IX{ }~XP_ALLXA1 3OSS0

图2
(BRJLDE)IVGLQ(IIF8P`DV8


欢迎您的提问。辛苦您提供以下信息,以方便我们快速定位和解决问题:
PaddleRS版本:PaddleRS-1.0.0
PaddlePaddle版本:PaddlePaddle 2.4.0
操作系统信息:Windows
Python版本号:Python3.7
CUDA/cuDNN版本:CUDA11.6/cuDNN 8.4.0

@cctx1 cctx1 added the question Further information is requested label Apr 13, 2024
@github-actions github-actions bot added triage new issue/PR waiting to be dealed and removed triage new issue/PR waiting to be dealed labels Apr 13, 2024
@cctx1 cctx1 changed the title [General Issue] 目标检测预训练模型问题 Apr 13, 2024
@Bobholamovic
Copy link
Member

Bobholamovic commented Apr 15, 2024

由于预训练使用的数据集和当前模型训练使用的数据集不同,所以模型的head部分结构不同,这通常是可以接受的~ 从日志上来看,backbone部分的预训练权重已经得到成功加载。相比从头训练,模型应该能够享受使用预训练权重的好处。

@cctx1
Copy link
Author

cctx1 commented Apr 15, 2024

谢谢开发者回复,问题已经解决!

Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment
Labels
question Further information is requested
Projects
None yet
Development

No branches or pull requests

2 participants