Jiaming Han

是的

猜测pytorch，cuda不适配

This script may help you. ``` from mmdet.models import build_detector from mmcv.runner import load_checkpoint import numpy as np from mmdet.datasets import to_tensor from mmdet.datasets.transforms import ImageTransform import mmcv import argparse...

缩放至相同尺寸后相加

不同fpn层的feature map resize到一样大

Please refer to https://github.com/dingjiansw101/AerialDetection

Sorry for that, I do not have the checkpoints of these models. Please raise issues in https://github.com/dingjiansw101/AerialDetection

需要

I guess the model is overfitted to your training data. If you only have a few images, you can use our pretrained model on DOTA and fine-tune on your own...

https://github.com/csuhan/ReDet/blob/master/GETTING_STARTED.md