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

The models 'r2plus1d_34_clip32_ft_kinetics_from_ig65m' fine-tuned on UCF101 can't get good performance, just 78%, why? #28

Closed
fwcheng opened this issue Oct 19, 2019 · 14 comments

Comments

@fwcheng
Copy link

fwcheng commented Oct 19, 2019

Thanks for your models. I met with some problem when using the model 'IG-65M+KINETICS+32X112X112'. I fine-tuned on UCF101 dataset using it, but the result is 78% approximately, which is far away from the results in the paper. Did you get the same results of paper's author? If so, can you share more details about your experiments. Thanks very much!

@fwcheng fwcheng changed the title The The models 'r2plus1d_34_clip32_ft_kinetics_from_ig65m' fine-tuned on UCF101 can't get good performance, just 78%, why? Oct 19, 2019
@daniel-j-h
Copy link
Member

I don't know your fine-tuning protocol, code, or methodology - so can't say anything about your specific results. The official caffe2 models are here if you want to try them instead:

https://github.com/facebookresearch/vmz

We have a ticket open for replicating the official results on Kinetics (see #2) but it's currently blocked by Kinetics being impossible to get our hands on.

@sandhawalia
Copy link
Member

@fwcheng - Can you share your fine-tuning code/repo with us.

@daniel-j-h
Copy link
Member

@fwcheng any updates here?

@fwcheng
Copy link
Author

fwcheng commented Nov 2, 2019

@fwcheng any updates here?

Sorry for the late reply. I have solved this problem, and the model parameters you provided are useful.

@fwcheng
Copy link
Author

fwcheng commented Nov 2, 2019

@fwcheng - Can you share your fine-tuning code/repo with us.

I haven‘t finished my project, so I didn't share it on Github.

@daniel-j-h
Copy link
Member

Great you could solve the problem on your end! Closing here as not actionable on our side.

@DAVEISHAN
Copy link

Hey @fwcheng nice to hear that you got your problem solved. How much accuracy are you getting on UCF-101 now? Did you make any changes in the model/ pretrained weights to achieve that?

Eagerly waiting for your reply.

Regards,
Ishan

@fwcheng
Copy link
Author

fwcheng commented Dec 25, 2019

Hey @fwcheng nice to hear that you got your problem solved. How much accuracy are you getting on UCF-101 now? Did you make any changes in the model/ pretrained weights to achieve that?

Eagerly waiting for your reply.

Regards,
Ishan

Sorry for the late reply. The results I got on UCF101 is 96.8%, and I haven't change the pretrained weights, I just changed the learning rate for finetune.

@Madara321
Copy link

Hi @fwcheng I also got about 80% on UCF101, could you share more details about finetune?

@FesianXu
Copy link

@fwcheng Hi, to get 96.8% on UCF101, how many clips did you use ?

@Yueeeeee-1
Copy link

hi, same problem i've met as you met before, the result is only 78.5% when I fine-tuned on UCF101, can you show more details about the learning rate setting, and can you share the way of updating learning rate? Thank you!

@paden118
Copy link

paden118 commented Dec 9, 2021

hi, same problem i've met as you met before, the result is only 78.5% when I fine-tuned on UCF101, can you show more details about the learning rate setting, and can you share the way of updating learning rate? Thank you!

Hi, what is the accuracy of your final fine-tuning on ucf101?

@fwcheng
Copy link
Author

fwcheng commented Dec 9, 2021 via email

@paden118
Copy link

paden118 commented Dec 9, 2021

你好,请问你是微调所有层吗?学习率是如何设置的呢?比如所有卷积层和最后一层全连接层的学习率是分别设置多少?谢谢

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

8 participants