OW-DETR
OW-DETR copied to clipboard
[CVPR 2022] Official Pytorch code for OW-DETR: Open-world Detection Transformer
Hello! How to trian "19+1" , "15+5" and "10+10" settings?
In this setting, can you send me the trained model? Because there is a gap between my results and the original.
Dear Author, In the paper, I see that every task is trained for 50 epochs and finetuned for 20 epochs as  However, in `configs/OWOD_new_split.sh`, I see the training schedule...
In Task1, my current class AP50 was 73.0, which was quite different from the results in the paper, and the U-recall was only 4.4. I don't know why. Can you...
I didnt find the objectness related calculation in the code, could you kindly point it out ? THX
how to train old owod split?
Hello, when training with OWOD proposed splits, why are there 5 '0.0' in AP50? **Task 1** "AP50 ": "['64.0', '32.8', '30.0', '26.5', '57.1', '38.1', '78.3', '43.1', '60.7', '45.9', '46.3', '42.9',...
can not download MultiScaleDeformableAttention by using pip
Hi, So I have been trying to implement the old TOWOD data split. I found that the images specified there are the VOC2007 classes. However, there is a bug where...