TypeError: can't pickle _thread.RLock objects

[kelek2]

Hi:
when we use your tool img_split.py, meet the error TypeError: can't pickle _thread.RLock objects,
happens " patch_infos = pool.map(worker, zip(infos, img_dirs))" , looks like multiprocessing problem , can you suggest us how to solve the problem， thanks

[jbwang1997]

I haven't met this error in my python 3.7.3. I recommend you updating the vision of python or using one processing by setting --nproc 1.

[kelek2]

Thank you for your reply, and I would like to ask again. When using FASTER RCNN OBB to train the DOTA dataset, first I want to use the img_split.py tool to cut the image, and whether the input dota txt annotation data format is (x1, x2, y1, y2, x3 ,y3,x4,y4,lable,diff)

[jbwang1997]

Yes. img_split.py can directly load DOTA annotations by setting --load_type dota. There are also other annotation formats for different datasets.

[kelek2]

Thanks, I use img_split.py split the image and make the annotation files, but when we use FASTER RCNN OBB method train the splited DOTA dataset, run tools/train.py ,
we meet the problem "ValueError: operands could not be broadcast together with shapes (3,8) (4,) ",
where it happed in the "mmdet/datasdets/pipelines/transforms.py",
function" def _resize_bboxes(self, results):
"""Resize bounding boxes with results['scale_factor']."""
img_shape = results['img_shape']
for key in results.get('bbox_fields', []):
bboxes = results[key] * results['scale_factor']
bboxes[:, 0::2] = np.clip(bboxes[:, 0::2], 0, img_shape[1])
bboxes[:, 1::2] = np.clip(bboxes[:, 1::2], 0, img_shape[0])
results[key] = bboxes"
we think it the train image or annotations format problem, can u give us a brief introdaction how to make the data format to train the DOTA datesets using the FASTER RCNN OBB, the config file is "configs/obb/faster_rcnn_obb/faster_rcnn_obb_r50_fpn_1x_dota10.py", thanks!

[jbwang1997]

Sorry, this is the first time I meet this error, You can check oriented_model_starting for details of using OBBDtection.

[XuJieLing-cloud]

@kelek2 have you solved your problem? i am confused , too