Skip to content

Is res50h pertained on ImageNet? #8

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

Open
shuzhangcasia opened this issue Sep 12, 2019 · 2 comments
Open

Is res50h pertained on ImageNet? #8

shuzhangcasia opened this issue Sep 12, 2019 · 2 comments

Comments

@shuzhangcasia
Copy link

In your experiment using res50h model, do you use image-net pretrained weights or randomly initialized weights?

@twangnh
Copy link
Owner

twangnh commented Jan 16, 2020

yes, the backbone is pretrained with imagenet

@inverse-limit
Copy link

How did you set the hyperparameter when training the baseline res50 model? I apply your method to SSD and it works. But in faster-rcnn case, I replace the vgg in your code with pretrained resnet50 and distill it with a resnet101 model on VOC2007 for 7 epoch, batch-size 1, 0.001 lr decay 1/10 at epoch 5. The final mAP of res50 model is 72.2 which is close to your paper. But without kd, the mAP can reach 74.0.

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

3 participants