cannot import name 'sphere_synt' from partially initialized module 'ref'

When I run det/yolov4/train_yolov4.sh configs/yolov4/bop_pbr/yolov4x_640_augCosyAAEGray_ranger_lm_pbr_lm_test.py 0, It produces an error. I found that there are no “sphere_synt.py” files under the “def” directory.How can I fix it?

It is not needed for this repo. You can just comment it out.

Checkpoint pretrained_models/yolov4/yolov4x-mish.pth not found!Now I have this problem.Please how can I solve it?

https://github.com/WongKinYiu/PyTorch_YOLOv4

How do I use it？I click on the "weights" of the pre-trained model inside, but it shows that the file does not exist.

Prior to this,when I run sh scripts/compile. sh command, it displays the following content.

May I ask if this has any impact?

Looks like the egl renderer could not be successfully built in your env. You need to fix the problem.

Yes, how can I fix this problem？

My env has already been installed PyOpenGL

Maybe you can refer to: shanice-l/gdrnpp_bop2022#35 and some other related issues online.

Thank you very much for your help! I have resolved the egl renderer issue, but this(#16 (comment)) still exists.

I entered this(https://github.com/WongKinYiu/PyTorch_YOLOv4), but its weights file no longer exists

Are these different profiles different datasets? What is the difference between config inside the base folder and the four outside?

Yolov4's processing on LM I see only test.py. So is the reason for my error(#16 (comment)) being because I run det/yolov4/train_yolov4.sh and let configs set the yolov4x_640_augCosyAAEGray_ranger_lm_pbr_lm_test?

Please find the answer in the code https://github.com/THU-DA-6D-Pose-Group/self6dpp/blob/main/det/yolov4/main_yolov4.py

This looks like the following code is causing problems, but I don't know how to properly modify it to make it correct

self6dpp/det/yolov4/main_yolov4.py

Line 181 in d94eaf7

launch(

Can you provide a pretrained_ models of yolov4x-mish.pth? Because I found that its download link has expired in its Yolov4 GitHub.

I don't have the yolov4 models now. Please contact the original authors.

Thanks to your help, I am now able to train The Detector.

Can I adjust the TOTAL_EPOCHS=4 here to make it less training? Because I find that training the detector takes too long, I use a piece of 2080ti, will this have an impact beyond accuracy?

You can try it and see how much it impacts the accuracy.