AerialDetection

AerialDetection copied to clipboard

ModuleNotFoundError: No module named 'mmdet.core.bbox.bbox'

6

Traceback (most recent call last): File "tools/train.py", line 7, in from mmdet.datasets import get_dataset File "/home/lab716/xukaiyan/AerialDetection/mmdet/datasets/__init__.py", line 1, in from .custom import CustomDataset File "/home/lab716/xukaiyan/AerialDetection/mmdet/datasets/custom.py", line 11, in from .extra_aug...

xukaiyan

tools/test.py中似乎不支持map的计算，只有bbox map的计算，请问这个repo怎么计算map？

9

mmdzzh

Hi @dingjiansw101 , What license is DOTA_Devkit under? I've raised the question [here](https://github.com/CAPTAIN-WHU/DOTA_devkit/issues/48) as well

ashnair1

IndexError: string index out of range

8

你好： 我在自己的数据集上进行了训练，也生成了测试结果文件results.pkl。然而在使用parse_results.py进行分析的时候出现了下面问题。 Traceback (most recent call last): File "tools/parse_results.py", line 129, in parse_results(config_file, pkl_file, output_path, type) File "tools/parse_results.py", line 67, in parse_results obb_results_dict = OBBDetComp4(dataset, outputs) File "/dat01/baijing/wenzheng/AerialDetection/mmdet/core/evaluation/dota_utils.py", line 188,...

wennzheng

Convert high resolution images to 1024 * 1024 image patch (multiple) while also augmenting the annotation.

1

How to use the DOTA_devkit shared by you to do this image split?

Karim-Baig

请问，这个检测旋转框，是直接用原始的文本标签嘛？还是在本工程里转为了coco格式

1

18804601171

DOAI1_5 performance is too low for the baseline model

4

Thanks for sharing the code! I have tested the repo on DOAI 1.5 and submit it to the online evaluation task 1. But the result is too low. ``` mAP:...

tbuikr

18804601171

mask转poly中的异常

5

您好，我在实验过程中，一开始将dota原图像剪切为1024*1024，gap512，然后训练模型，没有问题。但是当我重新剪切了一个600*600，gap150的数据集，并用同样的模型训练，训练时scale为800*800时，在transforms_rbbox.py的mask2poly_single中抛出了异常： /mmdet/core/bbox/transforms_rbbox.py(577)mask2poly_single() -> return poly (Pdb) 似乎是mask的值有问题导致进入这个异常。这是剪切图像的问题还是什么其他的的原因么？请问我应该如何处理？

KKKiven

If I want to add some anchor default angles or modify them? Thanks for reply! 大佬们怎么修改默认旋转锚框的旋转角

tjmannn