TDD icon indicating copy to clipboard operation
TDD copied to clipboard

Published 20 hours ago • verified publisher icon Feobi1999

Can you provide google drive link for pretrained models please?

Open tysnow2022 opened this issue 2 years ago • 6 comments

tysnow2022 avatar May 13 '22 22:05 tysnow2022

Ok, I will provide the google link in this week. Btw, the data and environment preparation will be updated this weekend.

Feobi1999 avatar May 16 '22 12:05 Feobi1999

Oh thankyou glad to hear that. I was struggling by my own. actually for the pretrained model a friend with Baidu account helped me. When i try to evaluate it i got the following issue. I attached the screen shoot of the error and for full log information here is the link. https://drive.google.com/file/d/1avtwHQV9TQoRA-n7vOOANsvGkdoITVzW/view?usp=sharing log1

tysnow2022 avatar May 16 '22 13:05 tysnow2022

@tysnow2022 When I deleted some redundant code, I deleted a line of content by mistake. I just fixed it and tested it, you can pull the latest version code. Sorry for the inconvenience.

Feobi1999 avatar May 16 '22 14:05 Feobi1999

thank you the testing works but the training throws this error. error3

tysnow2022 avatar May 16 '22 15:05 tysnow2022

The training works well in my experiment server. Mention now we only support the one gpu one pair images strategy, which has been introduced in the Readme file.

Feobi1999 avatar May 16 '22 16:05 Feobi1999

I am replying so lately. sorry for the inconvenience. I set the IMG_PER_BATCH_LABEL and IMG_PER_BATCH_UNLABEL both to 1 in the config yaml file . and it starts the training but it shows the loss is 'nan' like this "total_loss: nan, loss_cls: nan loss_box_reg: nan loss_cls_head_2: nan"

tysnow2022 avatar May 17 '22 16:05 tysnow2022