Kaggle cryo #1947
Kaggle cryo #1947
Conversation
for more information, see https://pre-commit.ci
DCO Remediation Commit for christofhenkel <[email protected]> I, christofhenkel <[email protected]>, hereby add my Signed-off-by to this commit: 1db03fd I, christofhenkel <[email protected]>, hereby add my Signed-off-by to this commit: 2ffe23d I, christofhenkel <[email protected]>, hereby add my Signed-off-by to this commit: df958e7 Signed-off-by: christofhenkel <[email protected]> Signed-off-by: ChristofHenkel <[email protected]>
Signed-off-by: ChristofHenkel <[email protected]>
Signed-off-by: ChristofHenkel <[email protected]>
Signed-off-by: ChristofHenkel <[email protected]>
There was a problem hiding this comment.
Hi @ChristofHenkel,
thank you for your contribution and congratulations on your great achievement.
I am trying to follow your README.md file instructions to run the code myself, but I have encountered some issues.
I leave you some comments. Please let me know if I am on the right track.
Many thanks,
Rafael
| This tutorial is build upon the official Cryo ET competition data. It can be downloaded directly from kaggle: https://www.kaggle.com/competitions/czii-cryo-et-object-identification/data | ||
|
|
||
| Alternativly it can be downloaded using the kaggle API (which can be installed via ```pip install kaggle```) | ||
|
|
There was a problem hiding this comment.
If you decide to use the Kaggle API you need to create a Kaggle account and configure your token as described here.
There was a problem hiding this comment.
You will also need to follow the competition url and click "join competition" to accept the terms and conditions and then be allowed to download the data with the following command:
| def load_one(self, experiment_id): | ||
|
|
||
| img_fp = f"{self.data_folder}{experiment_id}" | ||
| try: |
There was a problem hiding this comment.
This try statement will defer the error to line 88, where img will not be defined. Consider replacing the print statement with a raise or instantiating img on the exception.
There was a problem hiding this comment.
While debugging the code, I found that it was erroring out because of this error:
File "/workspace/data/ds_1.py", line 83, in load_one
img = np.array(zarr.open(img_fp + '/VoxelSpacing10.000/denoised.zarr')[0]).transpose(2,1,0)
File "/usr/local/lib/python3.10/dist-packages/zarr/hierarchy.py", line 511, in __getitem__
raise KeyError(item)I confirmed the img_fp + '/VoxelSpacing10.000/denoised.zarr' path exists and it does, but zf is empty:
> zf
array([], dtype=float64)I may have downloaded the data incorrectly. @ChristofHenkel, could you please take a look at this?
The contents of my folder are these:
$ ls /data/train/static/ExperimentRuns/TS_86_3/TS_86_3/VoxelSpacing10.000/denoised.zarr/.zgroup -la
-rw-r--r-- 1 root root 24 Mar 7 17:06 /data/train/static/ExperimentRuns/TS_86_3/TS_86_3/VoxelSpacing10.000/denoised.zarr/.zgroup|
|
||
| Alternativly it can be downloaded using the kaggle API (which can be installed via ```pip install kaggle```) | ||
|
|
||
| ```kaggle competitions download -c czii-cryo-et-object-identification``` |
There was a problem hiding this comment.
If you do this inside of the docker pod, this data will be ephemeral and be deleted when the pod goes down. I suggest adding an option to mount the data in the correct location and instruct people to download the data outside the pod.
|
|
||
| ```kaggle competitions download -c czii-cryo-et-object-identification``` | ||
|
|
||
| and adjust path to it in ```configs/common_config.py``` with ```cfg.data_folder```. |
There was a problem hiding this comment.
If you mount the folder in the correct location you don't need to adjust this. So this line can be removed.
Description
Added tutorial based on the 1st place solution of kaggles Cryo-ET competition