trying to use polarmask but there is no config for single gpu!

[bat3a]

trying to use Polarmask but there is no config for single gpu! there is only for 4, 32 gpu in the config folder, is there away to test single gpu?

[xieenze]

hi
PolarMask is just based on mmdetection. You can find some examples in https://github.com/open-mmlab/mmdetection/blob/v1.0rc0/GETTING_STARTED.md.

For example, if you want to test on single gpu or multi-gpu , you can try this

# single-gpu testing
python tools/test.py ${CONFIG_FILE} ${CHECKPOINT_FILE} [--out ${RESULT_FILE}] [--eval ${EVAL_METRICS}] [--show]

# multi-gpu testing
./tools/dist_test.sh ${CONFIG_FILE} ${CHECKPOINT_FILE} ${GPU_NUM} [--out ${RESULT_FILE}] [--eval ${EVAL_METRICS}]

[bat3a]

thanks for the fast response, i already modified the config and downloaded the retrained model:

config_file = '../configs/polarmask/4gpu/polar_768_1x_r101.py'
checkpoint_file = '../work_dirs/r101_1x.pth'

but i am trying to run the visualize jupyter notebook code, i get the following error:
KeyError: 'PolarMask is not in the detector registry'

i verfied i can import mmdet, any help would be appreciated.

[ZhongYingMatrix]

@bat3a git pull and remake?

[bat3a]

@ZhongYingMatrix for what repo, mmcv, mmdet?! both are latest

[xieenze]

@bat3a
Sorry, I don't know what's wrong with you. Maybe you can refer to this issue.

The only potential problem is that I found I changed the name of convs in the later version,
but the published model is out-dated.
So I upload the new model in google drive and I test it can work.

You can re-download the code and the trained model to test it again. I think you can successfully run demo in this way.

[bat3a]

@xieenze i have installed all the necessary libs and can run the demos in the mmdetection successfully,
but when i try to run the notebook in the polarmask/demo i get this error:

what should i do to register the polarmask with mmdet?! thanks in advance.

[xieenze]

@bat3a
Do you follow my advice that re-download the code and the model？

[bat3a]

@bat3a
Do you follow my advice that re-download the code and the model？

yes the code is up to date and i redownloaded the .pth file today.

[bat3a]

if it helps my folders arranged like this:
Desktop
|------mmdetection
|------Polarmask

also i use ubuntu 18.04, cuda 9.2.

[bat3a]

@xieenze can you tell me if i mmdetection should be installed? as per instructions, as there is an mmdet folder of the lib in the polarmask folder and i think it conflicts with the installation!

[xieenze]

@bat3a
You do not need to download mmdetection.
You only need to download PolarMask and install, same as mmdetection.

[bat3a]

thanks that worked
the problem was in the installation instruction, it says clearly that you need to install mmdetection which is confusing!

[xieenze]

@bat3a
Hello,
Sorry to make you confuse. I just want to say the install process is same as MMDetection.........=-=

[wufan-zhao]

@xieenze hi Enze, fabulous work! However, I met the same issue as @bat3a did when running the demo notebook. I know this project is built based on mmdetection, but the instruction for installing and configuration is a bit confusing indeed. I would suggest re-write an independent installing instruction just for polarmask.
Now I have built two separate envs for mmdetection and polarmask separately, but I encountered another error:

---

ModuleNotFoundError Traceback (most recent call last)
in
----> 1 from mmdet.apis import init_detector, inference_detector, show_result_pyplot
2 import mmcv

~/Projects/PolarMask/mmdet/apis/init.py in
1 from .env import get_root_logger, init_dist, set_random_seed
----> 2 from .inference import (inference_detector, init_detector, show_result,
3 show_result_pyplot)
4 from .train import train_detector
5

~/Projects/PolarMask/mmdet/apis/inference.py in
8 from mmcv.runner import load_checkpoint
9
---> 10 from mmdet.core import get_classes
11 from mmdet.datasets import to_tensor
12 from mmdet.datasets.transforms import ImageTransform

~/Projects/PolarMask/mmdet/core/init.py in
1 from .anchor import * # noqa: F401, F403
2 from .bbox import * # noqa: F401, F403
----> 3 from .evaluation import * # noqa: F401, F403
4 from .fp16 import * # noqa: F401, F403
5 from .mask import * # noqa: F401, F403

~/Projects/PolarMask/mmdet/core/evaluation/init.py in
3 voc_classes)
4 from .coco_utils import coco_eval, fast_eval_recall, results2json
----> 5 from .eval_hooks import (CocoDistEvalmAPHook, CocoDistEvalRecallHook,
6 DistEvalHook, DistEvalmAPHook)
7 from .mean_ap import average_precision, eval_map, print_map_summary

~/Projects/PolarMask/mmdet/core/evaluation/eval_hooks.py in
11 from torch.utils.data import Dataset
12
---> 13 from mmdet import datasets
14 from .coco_utils import fast_eval_recall, results2json
15 from .mean_ap import eval_map

~/Projects/PolarMask/mmdet/datasets/init.py in
1 from .builder import build_dataset
----> 2 from .cityscapes import CityscapesDataset
3 from .coco import CocoDataset
4 from .custom import CustomDataset
5 from .dataset_wrappers import ConcatDataset, RepeatDataset

~/Projects/PolarMask/mmdet/datasets/cityscapes.py in
1 from .coco import CocoDataset
2 from .registry import DATASETS
----> 3 from .coco_seg import Coco_Seg_Dataset
4
5

~/Projects/PolarMask/mmdet/datasets/coco_seg.py in
6 import cv2
7 import math
----> 8 import Polygon as plg
9 from tqdm import tqdm
10

ModuleNotFoundError: No module named 'Polygon'.
The mmdetection runs well, but not this. what would be the potential cause of this error?

Second, I would also like to ask "how can I retrain the model on single gpu?" many thanks

[LittleQQ]

thanks for the fast response, i already modified the config and downloaded the retrained model:

config_file = '../configs/polarmask/4gpu/polar_768_1x_r101.py'
checkpoint_file = '../work_dirs/r101_1x.pth'

but i am trying to run the visualize jupyter notebook code, i get the following error:
KeyError: 'PolarMask is not in the detector registry'

i verfied i can import mmdet, any help would be appreciated.

hi
PolarMask is just based on mmdetection. You can find some examples in https://github.com/open-mmlab/mmdetection/blob/v1.0rc0/GETTING_STARTED.md.

For example, if you want to test on single gpu or multi-gpu , you can try this

# single-gpu testing
python tools/test.py ${CONFIG_FILE} ${CHECKPOINT_FILE} [--out ${RESULT_FILE}] [--eval ${EVAL_METRICS}] [--show]

# multi-gpu testing
./tools/dist_test.sh ${CONFIG_FILE} ${CHECKPOINT_FILE} ${GPU_NUM} [--out ${RESULT_FILE}] [--eval ${EVAL_METRICS}]

i has same problem , Have you solved the problem now？expect your answer