Can't import a COCO Dataset

[nasserdr]

I have one image and one annotation file that I am trying to import using datumaro but I am getting an error:

could't match this file with any of these tasks: coco_instances,coco_person_keypoints ...

My code is:
`from datumaro import Dataset

dataset = Dataset.import_from("/home/dataset", format="coco")

dataset
`

The dataset folder is structured as follows:
home
----dataset
--------Images
------------DJI_20210901143844_0969_02_02.png
--------Annotations
------------DJI_20210901143844_0969_02_02.json

I am getting this error:

`---------------------------------------------------------------------------
DatasetNotFoundError Traceback (most recent call last)
Cell In[10], line 13
8 ids += [item.id]
10 return ids
---> 13 dataset = Dataset.import_from("/home/naro/data/RumexV0/Test2", format="coco")
14 subset = "default"
15 ids = get_ids(dataset, subset)

File ~/.conda/envs/images/lib/python3.10/site-packages/datumaro/components/dataset.py:832, in Dataset.import_from(cls, path, format, env, progress_reporter, error_policy, **kwargs)
829 importer = env.make_importer(format)
830 with logging_disabled(log.INFO):
831 detected_sources = (
--> 832 importer(path, stream=cls._stream, **kwargs)
833 if importer.can_stream
834 else importer(path, **kwargs)
835 )
836 extractor_merger = importer.get_extractor_merger()
837 elif format in env.extractors:

File ~/.conda/envs/images/lib/python3.10/site-packages/datumaro/plugins/data_formats/coco/importer.py:78, in CocoImporter.call(self, path, stream, **extra_params)
75 subsets = self.find_sources(path)
77 if len(subsets) == 0:
---> 78 raise DatasetNotFoundError(path, self.NAME)
80 # TODO: should be removed when proper label merging is implemented
...
(...)
86 CocoTask.stuff,
87 }

DatasetNotFoundError: Failed to find dataset 'coco' at '/home/dataset'
`

Attached the json file and the image:
DJI_20210901143828_0957_02_02.json
(I can't upload the image because it's more than 10 MB)

[sooahleex]

Hi @nasserdr sorry for the late reply, thanks for your interest in our project.
I checked the structure of your dataset, and your dataset has a little bit different structure with which we support. This is the structure we support for coco

└─ Dataset/
    ├── dataset_meta.json # a list of custom labels (optional)
    ├── images/
    │   ├── train/
    │   │   ├── <image_name1.ext>
    │   │   ├── <image_name2.ext>
    │   │   └── ...
    │   └── val/
    │       ├── <image_name1.ext>
    │       ├── <image_name2.ext>
    │       └── ...
    └── annotations/
        ├── <task>_<subset_name>.json
        └── ...

For more information, please refer to this document.
So your dataset does not have a subset for each image. Please check the structure of the dataset and if there is a problem even though it is the same as the structure we support, please give me an example of the dataset and let me know and I will check it.