v2.4.1 raises ValueError: not enough values to unpack when calling predict_single_npy_array

[joshuacwnewton]

Note

I see a flurry of activity on the PyPI History, so it's possible that you're already aware of this. If so, my apologies for adding to the noise!

• We have a CI workflow that installs the most up-to-date version of nnuntv2 and runs inference tests.
• Our CI started failing after the upgrade to v2.4. Retrying the tests with v2.4.1 results in the same error.

Traceback:

  File "/__w/spinalcordtoolbox/spinalcordtoolbox/spinalcordtoolbox/deepseg/inference.py", line 228, in segment_nnunet
    pred = predictor.predict_single_npy_array(
  File "/__w/spinalcordtoolbox/spinalcordtoolbox/python/envs/venv_sct/lib/python3.9/site-packages/nnunetv2/inference/predict_from_raw_data.py", line 442, in predict_single_npy_array
    dct = next(ppa)
  File "/__w/spinalcordtoolbox/spinalcordtoolbox/python/envs/venv_sct/lib/python3.9/site-packages/batchgenerators/dataloading/data_loader.py", line 126, in __next__
    return self.generate_train_batch()
  File "/__w/spinalcordtoolbox/spinalcordtoolbox/python/envs/venv_sct/lib/python3.9/site-packages/nnunetv2/inference/data_iterators.py", line 193, in generate_train_batch
    image, seg_prev_stage, props, ofname = self._data[idx][0]
ValueError: not enough values to unpack (expected 4, got 1)

nnUNet/nnunetv2/inference/data_iterators.py

Lines 191 to 193 in 9945333

	def generate_train_batch(self):
	idx = self.get_indices()[0]
	image, seg_prev_stage, props, ofname = self._data[idx][0]

It looks like this issue may be due to d87fa5b. (I don't think the [0] should be present anymore for the unpacking to work successfully.)

Thank you for your time and effort maintaining this package! ♥️

[joshuacwnewton]

I've opened a PR for this issue in #2114. :)

[FabianIsensee]

Thanks a lot! I am currently traveling and will merge it once I am back :-)