zeyang

Thank you for pointing out this issue. It is actually a redundant step. I substitute it with the manual instructions **Image shots and splits preparation** in `Readme.md` so that you...

Hi, sorry that we do not support input with size 512. However, you can consider to recover some lines of codes that we've deleted for simplicity from [RFBNet](https://github.com/ruinmessi/RFBNet) to make...

Thanks for your interest, I will update the results soon.

You can download our [dataset](https://drive.google.com/file/d/1X-ej4QQXf6UgpIz13nSpl__r3FjhvUx2/view?usp=sharing) to test the code first.

We have added visualization feature in `utils/visualization.py`, hope that it helps. Thanks.

By default, we first cut out the upper 1/3 part of the input image. Then we randomly resize the width of the image to 1x~1.2x of `img_width` with the aspect...

@thomasehuang I can get the similar split1 5shot results with the reported one, ~ 55% nAP50 with the base model you released. However, if I train the base model by...

I also get the same results with @GuangxingHan. For TFA w/ fc, I only get 34.86 while the paper reports 36.8. I indeed use the official codebase without any modification....

@GuangxingHan Have you try with the model the author released? I get the similar split1 5shot results with the reported one, ~ 55% mAP on novel classes. However, if I...

@mandal4 I indeed ran the code with 8 GPUs as the auther's default setting. Could you provide your hyperparams?