Dear chairs:
Hello! Our team has submitted dockers on September 21st. But we didn't receive the email until yesterday saying there is a problem with the dockers. But today is the deadline. Furthermore, this error is about GPU number. The previous email said that there is no need to set the GPU device number. But the error message prompts to set the GPU device number. I think this is not our fault. Now, I have submitted dockers again, but no one test it again! We are very anxious due to this situation. I want to know if the deadline is passed, the dockers are still have problems. What should we do? Can I submit it again?
I hope to receive feedback from the organizing committee as soon as possible. Thank you!
Created by Zhenliang Ni ZhenliangNi The deadline is now over. How should we resubmit our dockers? Dear @ZhenliangNi,
I tested now also your other docker (docker.synapse.org/syn20812106/mi_segmentation_v2) and here I could see that you are just using Stage_2.
Please take care that you always process all testing stages (Testing Stages != Challenge tasks)!
It is all explained in detail in our Challenge Description [[DOWNLOAD](https://www.synapse.org/Portal/filehandle?ownerId=syn18779624&ownerType=ENTITY&fileName=RobustMIS2019_Design.pdf&preview=false&wikiId=591266)]
**From our document:**
Point 22.b)
The performance assessment for the challenge will performed in three stages.
* Stage 1: The test data is taken from the procedures (patients) from which the training data were extracted.
* Stage 2: The test data is taken from the exact same type of surgery as the training data but from procedures (patients) not included in the training data.
* Stage 3: The test data is taken from a different but similar type of surgery (and different patients) compared to the training data.
[...]
Test data:
* Stage 1: 663 cases in total (325 cases for the proctocolectomy surgery and 338 cases for the rectal resection surgery)
* Stage 2: 514 cases in total (225 cases for the proctocolectomy surgery and 289 cases for the rectal resection surgery)
* Stage 3: 2880 cases for the UNKNOWN SURGERY
Dear ZhenliangNi,
```
RuntimeError: Attempting to deserialize object on CUDA device 2 but torch.cuda.device_count() is 2. Please us torch.load with map_location to map your storages to an existing devce.
```
In your previous version you checked how many GPUs are existing and than you tried to access one more than existed ([0, 1] existed, you tried to access [2]). That's why you got an error. So as mentioned in our mail to you, every docker will just get one GPU from our hardware cluster, thats why you should not manually set GPU numbers.
But now is your docker running with the following output.
```
Test: 99%|??????????| 426/429 [00:58<00:00, 7.29it/s]/opt/conda/lib/python3.6/site-packages/skimage/io/_io.py:141: UserWarning: /output/Stage_1/Prokto/7/6000/output.png is a low contrast image
warn('%s is a low contrast image' % fname)
Test: 100%|??????????| 427/429 [00:58<00:00, 7.32it/s]/opt/conda/lib/python3.6/site-packages/skimage/io/_io.py:141: UserWarning: /output/Stage_1/Prokto/7/25500/output.png is a low contrast image
warn('%s is a low contrast image' % fname)
Test: 100%|??????????| 428/429 [00:58<00:00, 7.40it/s]/opt/conda/lib/python3.6/site-packages/skimage/io/_io.py:141: UserWarning: /output/Stage_1/Prokto/7/91500/output.png is a low contrast image
warn('%s is a low contrast image' % fname)
Test: 100%|??????????| 429/429 [00:58<00:00, 7.19it/s] [***censored***]
stage 1 finished!
```
Please be aware to process all Teststages, not just Stage 1.
Good luck.
With best regards
Drop files to upload
If dockers have problem after the deadline. What should we do? page is loading…