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

Weights for yolov5 & training loss #21

Open
darkC00kie opened this issue Jan 12, 2022 · 3 comments
Open

Weights for yolov5 & training loss #21

darkC00kie opened this issue Jan 12, 2022 · 3 comments
Labels
question Further information is requested

Comments

@darkC00kie
Copy link

❔Looking for weights for yolov5 and how to minimize training loss

Are there existing weights which can be used ?
And I have seen that you linked the weights from yolov4, do you have any advice on how to use them with this model?

Additionally what would you consider a "good" loss ?
I currently have a loss of about 200 which seems a bit high and yields results wich are not that good.
How can i improve my loss ?
Thank you for your help!

@darkC00kie darkC00kie added the question Further information is requested label Jan 12, 2022
@wang-zhix
Copy link

I also encountered the same problem. My loss is about 120. Have you finally solved it?

@darkC00kie
Copy link
Author

Not yet, although even if the high loss still exists, the resluts look quite promisingto me.
Are you also working with a custom data set ? And if so - did you manage to generate new anchors ?

@whiterAutumn
Copy link

I'm train the yolo5m for 20 epoch with voc dataset, the latest loss about 200。When I run test.py with conf=0.3 and iou=0.1 ,the output is below:
pred: person 0.3414 0 -83 500 416
bird 0.3518 0 -83 500 416
gt: horse 54.0 50.0 285.0 262.0
label: voc2022/JPEGImages/2007_000332.jpg 54.0,50.0,285.0,262.0,12

is there some bug in the test.py? How to soloe it? Look forward to your reply@LongxingTan @darkC00kie

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

3 participants