[BUG] load annotations exception while deleting labels

spaced commented 5 months ago

Describe the problem

load annotations results in an exception

Code to reproduce issue

>>> gs =fo.load_dataset('test')
>>> gs.list_annotation_runs()
['test1', 'test12']
>>> gs.load_annotations('test12')
Downloading labels from Label Studio...
Download complete
Loading labels for field 'ground_truth'...
Traceback (most recent call last):
  File "/home/testuser/git/testrepo/model/venv/lib64/python3.11/site-packages/fiftyone/core/odm/database.py", line 807, in bulk_write
    coll.bulk_write(batch, ordered=ordered)
  File "/home/testuser/git/testrepo/model/venv/lib64/python3.11/site-packages/pymongo/_csot.py", line 107, in csot_wrapper
    return func(self, *args, **kwargs)
           ^^^^^^^^^^^^^^^^^^^^^^^^^^^
  File "/home/testuser/git/testrepo/model/venv/lib64/python3.11/site-packages/pymongo/collection.py", line 569, in bulk_write
    bulk_api_result = blk.execute(write_concern, session)
                      ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
  File "/home/testuser/git/testrepo/model/venv/lib64/python3.11/site-packages/pymongo/bulk.py", line 576, in execute
    return self.execute_command(generator, write_concern, session)
           ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
  File "/home/testuser/git/testrepo/model/venv/lib64/python3.11/site-packages/pymongo/bulk.py", line 446, in execute_command
    _raise_bulk_write_error(full_result)
  File "/home/testuser/git/testrepo/model/venv/lib64/python3.11/site-packages/pymongo/bulk.py", line 157, in _raise_bulk_write_error
    raise BulkWriteError(full_result)
pymongo.errors.BulkWriteError: batch op errors occurred, full error: {'writeErrors': [{'index': 0, 'code': 2, 'errmsg': 'Cannot apply $pull to a non-array value', 'op': {'q': {'ground_truth.detections': {'$exists': True}}, 'u': {'$pull': {'ground_truth.detections': {'_id': {'$in': [ObjectId('65578a3b7499471472bb40e1'), ObjectId('6551158dfd0a50d1decf13be'), ObjectId('65bd13ddb36066900d2c2d9b'), ObjectId('65511598fd0a50d1decf3849'), ObjectId('6557445e59d3f439ba79cd5d'), ObjectId('65bd1531500399d9ed879671'), ObjectId('65577f7f66df65c20c6de81f'), ObjectId('65bd13ddb36066900d2c2cfc'), ObjectId('65bd13ddb36066900d2c2cfb'), ObjectId('65bd13ddb36066900d2c2cfd'), ObjectId('6551158dfd0a50d1decf13ba'), ObjectId('65bd13deb36066900d2c434a'), ObjectId('65bd13ddb36066900d2c2d0d')]}}}}, 'multi': True, 'upsert': False}}], 'writeConcernErrors': [], 'nInserted': 0, 'nUpserted': 0, 'nMatched': 0, 'nModified': 0, 'nRemoved': 0, 'upserted': []}

The above exception was the direct cause of the following exception:

Traceback (most recent call last):
  File "<stdin>", line 1, in <module>
  File "/home/testuser/git/testrepo/model/venv/lib64/python3.11/site-packages/fiftyone/core/collections.py", line 8837, in load_annotations
    return foua.load_annotations(
           ^^^^^^^^^^^^^^^^^^^^^^
  File "/home/testuser/git/testrepo/model/venv/lib64/python3.11/site-packages/fiftyone/utils/annotations.py", line 1079, in load_annotations
    _merge_labels(
  File "/home/testuser/git/testrepo/model/venv/lib64/python3.11/site-packages/fiftyone/utils/annotations.py", line 1486, in _merge_labels
    dataset.delete_labels(ids=_del_ids, fields=label_field)
  File "/home/testuser/git/testrepo/model/venv/lib64/python3.11/site-packages/fiftyone/core/dataset.py", line 3230, in delete_labels
    foo.bulk_write(sample_ops, self._sample_collection)
  File "/home/testuser/git/testrepo/model/venv/lib64/python3.11/site-packages/fiftyone/core/odm/database.py", line 816, in bulk_write
    raise ValueError(msg) from bwe
ValueError: Cannot apply $pull to a non-array value

System information

opensuse tumbleweed
python 3.11.8
FiftyOne v0.23.7
installed from pip

Willingness to contribute

The FiftyOne Community encourages bug fix contributions. Would you or another member of your organization be willing to contribute a fix for this bug to the FiftyOne codebase?

[ ] Yes. I can contribute a fix for this bug independently
[x] Yes. I would be willing to contribute a fix for this bug with guidance from the FiftyOne community
[ ] No. I cannot contribute a bug fix at this time

benjaminpkane commented 5 months ago

Hi @spaced, if you add reproduction steps we will be able to diagnose 🙏

andrewaf1 commented 1 month ago

I am experiencing this same issue as well. I loaded a detection dataset from a file, called annotate with classes=dataset.get_classes("gt_bounding_boxes") + ["small_blurry"] (the only difference from when I last did this), labeled 500 images in labels studio, and then called load_annotations.

voxel51 / fiftyone