ModuleNotFoundError: No module named 'point_rend' about lasr HOT 11 CLOSED

To predict segmentations, you need to clone detectron2 and replace path-to-detectron2-root as the root folder of detectron2 repo in this command.

Please see the updated instruction

from lasr.

Thanks for the reply.
I did as you said.
However, I get the same error.

I've also checked the torch and CUDA versions.
I have installed detectron2 correctly.
What could be the cause?

from lasr.

This line adds the point_rend module to the python path. Could you check whether 'path-to-detectron2-folder/projects/PointRend/' exists?

from lasr.

I updated the way of importing point_rend module to be independent of the detectron repo, but you still need to pass the repo path in order to read the model config files correctly.

from lasr.

'path-to-detectron2-folder/projects/PointRend/
does exist.

When I ran the updated code, I got the following error.
The file pointed to as non-existent by this error exists in the correct location.

Traceback (most recent call last):
File "mask.py", line 49, in
cfg.merge_from_file('%s/projects/PointRend/configs/InstanceSegmentation/pointrend_rcnn_X_101_32x8d_FPN_3x_coco.yaml'%(detbase))
File "/home/shiori/lasr-main/detectron2/detectron2/config/config.py", line 30, in merge_from_file
assert PathManager.isfile(cfg_filename), f"Config file '{cfg_filename}' does not exist!"
AssertionError: Config file './detectron2/projects/PointRend/configs/InstanceSegmentation/pointrend_rcnn_X_101_32x8d_FPN_3x_coco.yaml' does not exist!
/home/kana/lasr-main

from lasr.

Are you able to
ls ./detectron2/projects/PointRend/configs/InstanceSegmentation/pointrend_rcnn_X_101_32x8d_FPN_3x_coco.yaml?

I guess you may have cloned detectron2 to the root folder of lasr, but did not pass detectron2 dir relative to where the command is executed.

from lasr.

I couldn't run "ls".

I'm sorry.
I don't really understand what you mean, but where is the right place to clone it?

from lasr.

If you cloned detectron2 to ./lasr/detectron2, you could cd into ./preprocess, and then run
'python mask.py pika ../detectron2'

The point was to pass the relative path of detectron2 wrt the current folder to mask.py.

from lasr.

Thank you very much.
The problem has been solved.

By the way, where would I have cloned dentron2 to run with the original code?

Also, the first "log" and "tmp" files I created still do not contain anything, is that correct?

"python mask.py pika . After running "python mask.py pika ... /detectron2", the segmentation image is not generated.

from lasr.

It generates data in subfolders under lasr-root/database/DAVIS/

from lasr.

Thank you! The problem has been solved.

from lasr.