Jian Ding

The current code only support batch size=1.

Yes, of course you can. I have tried to apply RoITransformer on faster rcnn in mmdetection version. And it get even higher results. I will release the code later.

Have you changed the categories in your new dataset class?

We have this plan, but it is uncertain when it will be completed.

1. Yes, we do the multi-scale augmentation during the spliting. 2. The reason why we do not use the img_scale in mmdetection: (1) By using the img_scale, we need to...

@mmoghadam11 Can you share me more details? So that I can help you.

The RRPN results are from this repo https://github.com/DetectionTeamUCAS/RRPN_Faster-RCNN_Tensorflow.

1. Have you succeed in compiling the dota_kit? 2. To solve the error "PSROIALIGNAVEPooling", you need to compile the c++ operator as described in installation instructions.

Which model config you used? You could set the "MaxIoUAssignerCy". For example, https://github.com/dingjiansw101/AerialDetection/blob/a717a85953eab240435cbcbb39396481cd831068/configs/DOTA/faster_rcnn_obb_r50_fpn_1x_dota.py#L53 https://github.com/dingjiansw101/AerialDetection/blob/a717a85953eab240435cbcbb39396481cd831068/configs/DOTA/faster_rcnn_obb_r50_fpn_1x_dota.py#L76

You can rm the .so and other build files, and try to rebuild.