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

yolov8训练自己的数据集 #151

Closed
1 task done
iyangiii opened this issue Jun 1, 2023 · 8 comments
Closed
1 task done

yolov8训练自己的数据集 #151

iyangiii opened this issue Jun 1, 2023 · 8 comments
Assignees

Comments

@iyangiii
Copy link

iyangiii commented Jun 1, 2023

问题确认 Search before asking

  • 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

我在训练自己的数据集的过程中进行了eval 给出的数据是这样的 精度为0是为什么呢
image

@nemonameless
Copy link
Collaborator

#43
image

你好,请发下更多详细信息,包括配置文件并说明改动,以及此epoch的训练log,包括loss,可截图

@iyangiii
Copy link
Author

iyangiii commented Jun 2, 2023

好的 我没有用预训练权重
训练日志如下:
image
之前保留的vdl图片如下:
62d4118ede36c36405466621d2ae5e8
配置文件如下:(对数据集的路径和epoch进行了修改,因为我的数据集较小)
image
image
image
image
image
image

@iyangiii
Copy link
Author

iyangiii commented Jun 2, 2023

不好意思 配置文件图片发错了一张
image

@nemonameless
Copy link
Collaborator

加载coco权重预训练去试试吧,应该主要是这个原因。如果裸训才10epoch精度未必高得再等更多epoch。且总bs尽量不要低于32,资源不够用建议换小模型或改小尺度。
此外如果你需要改epoch数,则optimizer_500e.yml和yolov8_reader.yml里,所有的500epoch都得改成你设置的80epoch,reader里的mosaic_epoch也得改成相应70epoch。

@iyangiii
Copy link
Author

iyangiii commented Jun 2, 2023

加载coco权重预训练去试试吧,应该主要是这个原因。如果裸训才10epoch精度未必高得再等更多epoch。且总bs尽量不要低于32,资源不够用建议换小模型或改小尺度。 此外如果你需要改epoch数,则optimizer_500e.yml和yolov8_reader.yml里,所有的500epoch都得改成你设置的80epoch,reader里的mosaic_epoch也得改成相应70epoch。

已经按你说的操作了 bs也增大到了32 但是还是为0 加载过coco权重预训练了
image
image

@LDX17
Copy link

LDX17 commented Jul 10, 2023

@iyangiiiii 你好,请问你用的什么版本的paddle?我看你也是用的aistudio训练的yolov8,没有报错。我用过paddle2.4.1和2.5.0,都报错,无法训练。

@thanhhuyenpt
Copy link

@iyangiiiii I had the same problem, did you solve it?

@nemonameless
Copy link
Collaborator

从log图片看第二个epoch loss反而上升了,是不合理的,可能是数据集太小了,或者可以减小lr再试试。

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

4 participants