-
Notifications
You must be signed in to change notification settings - Fork 9
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
checkpoint and config_file #3
Comments
Hi, the reason of your problem is that you restored the weights of the base detector (PETR) for the tracker during inference. The tracker has more weights in the learnable data association module than the detector, which leads to the mismatch. You can still restore that checkpoint during training, in this case, all weights that matches the checkpoint are initialized from it, other weights are randomly initialized. However, during inference, you need to restore a checkpoint that exactly matches your model architecture. |
@dsx0511 A similar issue is observed during training. I download weights from PETR PETR-vov-p4-1600x640 & loaded it my config file. Is this expected?
|
Hello, thanks for your great work. I ran into a problem while running the ./tools/test.py using "/workspace/ADA-Track/plugin/configs/ada_track_petr.py /workspace/PF-Track/models/f1/f1_q5_fullres_e24.pth --seed 42 --eval map --out /workspace/ADA-Track/results".
![issue1](https://private-user-images.githubusercontent.com/84330888/344551351-6e42d213-9c60-43ba-9a73-8a868e93fd65.png?jwt=eyJhbGciOiJIUzI1NiIsInR5cCI6IkpXVCJ9.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.WbgWzYwoBh-9FAuep_nKnoY6ikewl70MOX86s7hPHg4)
The problem is "The model and loaded state dict do not match exactly". I download the checkpoint from PF-Track.
Is there any error in the usage of config file or any other errors present?Looking forward to your answer,thank you.
The text was updated successfully, but these errors were encountered: