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

使用自己的训练集train from scratch #193

Closed
EwardJohn opened this issue Jul 1, 2020 · 2 comments
Closed

使用自己的训练集train from scratch #193

EwardJohn opened this issue Jul 1, 2020 · 2 comments

Comments

@EwardJohn
Copy link

EwardJohn commented Jul 1, 2020

你好,我使用自己的训练集(只有1类物体)进行train from scratch ,但是训练的过程中,top1和top2始终是1.0000(eval也是这样的),如图所示:
使用的配置文件为resnet50_vd.yaml,在配置文件中我改了类别数为2,请问这种情况应该怎末更改配置文件?还有一个问题是,如何拿PaddleClas训练完成的分类模型使用PaddleDetection进行目标检测?谢谢!
image

@littletomatodonkey
Copy link
Collaborator

您好

  1. 您的train list和val list中包含了多少个类别呢?如果只包含一类的话 acc一直是1是正常的(但是这种情况也没有必要训练了)。
  2. 分类模型一般是作为PaddleDetection中的检测任务,如果您需要将PaddleClas中的预训练模型用在PaddleDetection中,需要将骨干网络集成在PaddleDetection中,然后再去搭建自己的网络。您可以参考PaddleClas中的vgg和PaddleDetection的vgg网络的写法
    https://github.com/PaddlePaddle/PaddleDetection/blob/master/ppdet/modeling/backbones/vgg.py
    https://github.com/PaddlePaddle/PaddleClas/blob/master/ppcls/modeling/architectures/vgg.py

@EwardJohn
Copy link
Author

好的,谢谢

HydrogenSulfate pushed a commit to HydrogenSulfate/PaddleClas that referenced this issue Oct 18, 2023
* add cylinder3d_unsteady_optimize ce

* add ce case

* Add ce

* Add ce
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment
Labels
None yet
Projects
None yet
Development

No branches or pull requests

2 participants