Datasets:

JyGuozzZ
/

MILL3D

Error code:   DatasetGenerationCastError
Exception:    DatasetGenerationCastError
Message:      An error occurred while generating the dataset

All the data files must have the same columns, but at some point there are 5 new columns ({'image_name', 'reference_image_path', 'reference_level_name', 'relative_image_path', 'split'}) and 2 missing columns ({'relative_image_dir', 'num_images'}).

This happened while the csv dataset builder was generating data using

hf://datasets/JyGuozzZ/MILL3D/metadata/samples.csv (at revision 0e0010554a1b2b18116d8824f61667d1abe88566), [/tmp/hf-datasets-cache/medium/datasets/89372289548924-config-parquet-and-info-JyGuozzZ-MILL3D-4171728f/hub/datasets--JyGuozzZ--MILL3D/snapshots/0e0010554a1b2b18116d8824f61667d1abe88566/metadata/levels.csv (origin=hf://datasets/JyGuozzZ/MILL3D@0e0010554a1b2b18116d8824f61667d1abe88566/metadata/levels.csv), /tmp/hf-datasets-cache/medium/datasets/89372289548924-config-parquet-and-info-JyGuozzZ-MILL3D-4171728f/hub/datasets--JyGuozzZ--MILL3D/snapshots/0e0010554a1b2b18116d8824f61667d1abe88566/metadata/samples.csv (origin=hf://datasets/JyGuozzZ/MILL3D@0e0010554a1b2b18116d8824f61667d1abe88566/metadata/samples.csv), /tmp/hf-datasets-cache/medium/datasets/89372289548924-config-parquet-and-info-JyGuozzZ-MILL3D-4171728f/hub/datasets--JyGuozzZ--MILL3D/snapshots/0e0010554a1b2b18116d8824f61667d1abe88566/metadata/scenes.csv (origin=hf://datasets/JyGuozzZ/MILL3D@0e0010554a1b2b18116d8824f61667d1abe88566/metadata/scenes.csv)]

Please either edit the data files to have matching columns, or separate them into different configurations (see docs at https://hf.co/docs/hub/datasets-manual-configuration#multiple-configurations)
Traceback:    Traceback (most recent call last):
                File "/usr/local/lib/python3.12/site-packages/datasets/builder.py", line 1800, in _prepare_split_single
                  writer.write_table(table)
                File "/usr/local/lib/python3.12/site-packages/datasets/arrow_writer.py", line 765, in write_table
                  self._write_table(pa_table, writer_batch_size=writer_batch_size)
                File "/usr/local/lib/python3.12/site-packages/datasets/arrow_writer.py", line 773, in _write_table
                  pa_table = table_cast(pa_table, self._schema)
                             ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
                File "/usr/local/lib/python3.12/site-packages/datasets/table.py", line 2321, in table_cast
                  return cast_table_to_schema(table, schema)
                         ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
                File "/usr/local/lib/python3.12/site-packages/datasets/table.py", line 2249, in cast_table_to_schema
                  raise CastError(
              datasets.table.CastError: Couldn't cast
              scene_name: string
              level_name: string
              capture_type: string
              exposure_value: double
              iso_value: double
              image_name: string
              relative_image_path: string
              reference_level_name: string
              reference_image_path: string
              split: string
              -- schema metadata --
              pandas: '{"index_columns": [{"kind": "range", "name": null, "start": 0, "' + 1522
              to
              {'scene_name': Value('string'), 'level_name': Value('string'), 'capture_type': Value('string'), 'exposure_value': Value('float64'), 'iso_value': Value('float64'), 'num_images': Value('int64'), 'relative_image_dir': Value('string')}
              because column names don't match
              
              During handling of the above exception, another exception occurred:
              
              Traceback (most recent call last):
                File "/src/services/worker/src/worker/job_runners/config/parquet_and_info.py", line 1347, in compute_config_parquet_and_info_response
                  parquet_operations = convert_to_parquet(builder)
                                       ^^^^^^^^^^^^^^^^^^^^^^^^^^^
                File "/src/services/worker/src/worker/job_runners/config/parquet_and_info.py", line 980, in convert_to_parquet
                  builder.download_and_prepare(
                File "/usr/local/lib/python3.12/site-packages/datasets/builder.py", line 882, in download_and_prepare
                  self._download_and_prepare(
                File "/usr/local/lib/python3.12/site-packages/datasets/builder.py", line 943, in _download_and_prepare
                  self._prepare_split(split_generator, **prepare_split_kwargs)
                File "/usr/local/lib/python3.12/site-packages/datasets/builder.py", line 1646, in _prepare_split
                  for job_id, done, content in self._prepare_split_single(
                                               ^^^^^^^^^^^^^^^^^^^^^^^^^^^
                File "/usr/local/lib/python3.12/site-packages/datasets/builder.py", line 1802, in _prepare_split_single
                  raise DatasetGenerationCastError.from_cast_error(
              datasets.exceptions.DatasetGenerationCastError: An error occurred while generating the dataset
              
              All the data files must have the same columns, but at some point there are 5 new columns ({'image_name', 'reference_image_path', 'reference_level_name', 'relative_image_path', 'split'}) and 2 missing columns ({'relative_image_dir', 'num_images'}).
              
              This happened while the csv dataset builder was generating data using
              
              hf://datasets/JyGuozzZ/MILL3D/metadata/samples.csv (at revision 0e0010554a1b2b18116d8824f61667d1abe88566), [/tmp/hf-datasets-cache/medium/datasets/89372289548924-config-parquet-and-info-JyGuozzZ-MILL3D-4171728f/hub/datasets--JyGuozzZ--MILL3D/snapshots/0e0010554a1b2b18116d8824f61667d1abe88566/metadata/levels.csv (origin=hf://datasets/JyGuozzZ/MILL3D@0e0010554a1b2b18116d8824f61667d1abe88566/metadata/levels.csv), /tmp/hf-datasets-cache/medium/datasets/89372289548924-config-parquet-and-info-JyGuozzZ-MILL3D-4171728f/hub/datasets--JyGuozzZ--MILL3D/snapshots/0e0010554a1b2b18116d8824f61667d1abe88566/metadata/samples.csv (origin=hf://datasets/JyGuozzZ/MILL3D@0e0010554a1b2b18116d8824f61667d1abe88566/metadata/samples.csv), /tmp/hf-datasets-cache/medium/datasets/89372289548924-config-parquet-and-info-JyGuozzZ-MILL3D-4171728f/hub/datasets--JyGuozzZ--MILL3D/snapshots/0e0010554a1b2b18116d8824f61667d1abe88566/metadata/scenes.csv (origin=hf://datasets/JyGuozzZ/MILL3D@0e0010554a1b2b18116d8824f61667d1abe88566/metadata/scenes.csv)]
              
              Please either edit the data files to have matching columns, or separate them into different configurations (see docs at https://hf.co/docs/hub/datasets-manual-configuration#multiple-configurations)

Need help to make the dataset viewer work? Make sure to review how to configure the dataset viewer, and open a discussion for direct support.

scene_name string	level_name string	capture_type string	exposure_value float64	iso_value float64	num_images int64	relative_image_dir string
01	Exp1_10_ISO3200	low_light	10	3,200	27	data/01/Exp1_10_ISO3200/images
01	Exp1_25_ISO3200	low_light	25	3,200	27	data/01/Exp1_25_ISO3200/images
01	GT	reference	null	null	27	data/01/GT/images
02	Exp1_15_ISO3200	low_light	15	3,200	26	data/02/Exp1_15_ISO3200/images
02	Exp1_5_ISO3200	low_light	5	3,200	26	data/02/Exp1_5_ISO3200/images
02	GT	reference	null	null	26	data/02/GT/images
03	Exp1_100_ISO3200	low_light	100	3,200	26	data/03/Exp1_100_ISO3200/images
03	Exp1_20_ISO3200	low_light	20	3,200	26	data/03/Exp1_20_ISO3200/images
03	Exp1_30_ISO3200	low_light	30	3,200	26	data/03/Exp1_30_ISO3200/images
03	GT	reference	null	null	26	data/03/GT/images
04	Exp1_10_ISO3200	low_light	10	3,200	25	data/04/Exp1_10_ISO3200/images
04	Exp1_15_ISO3200	low_light	15	3,200	25	data/04/Exp1_15_ISO3200/images
04	Exp1_40_ISO3200	low_light	40	3,200	25	data/04/Exp1_40_ISO3200/images
04	GT	reference	null	null	25	data/04/GT/images
05	Exp1_10_ISO3200	low_light	10	3,200	25	data/05/Exp1_10_ISO3200/images
05	Exp1_20_ISO3200	low_light	20	3,200	25	data/05/Exp1_20_ISO3200/images
05	Exp1_5_ISO3200	low_light	5	3,200	25	data/05/Exp1_5_ISO3200/images
05	GT	reference	null	null	25	data/05/GT/images
06	Exp1_30_ISO3200	low_light	30	3,200	24	data/06/Exp1_30_ISO3200/images
06	GT	reference	null	null	24	data/06/GT/images
07	Exp1_10_ISO3200	low_light	10	3,200	24	data/07/Exp1_10_ISO3200/images
07	Exp1_25_ISO3200	low_light	25	3,200	24	data/07/Exp1_25_ISO3200/images
07	GT	reference	null	null	24	data/07/GT/images
08	Exp1_10_ISO3200	low_light	10	3,200	23	data/08/Exp1_10_ISO3200/images
08	GT	reference	null	null	23	data/08/GT/images
09	Exp1_6_ISO3200	low_light	6	3,200	24	data/09/Exp1_6_ISO3200/images
09	GT	reference	null	null	24	data/09/GT/images
10	Exp1_10_ISO3200	low_light	10	3,200	24	data/10/Exp1_10_ISO3200/images
10	GT	reference	null	null	24	data/10/GT/images
indoor01	Exp1_125_ISO3200	low_light	125	3,200	24	data/indoor01/Exp1_125_ISO3200/images
indoor01	Exp1_50_ISO3200	low_light	50	3,200	24	data/indoor01/Exp1_50_ISO3200/images
indoor01	Exp1_80_ISO3200	low_light	80	3,200	24	data/indoor01/Exp1_80_ISO3200/images
indoor01	GT	reference	null	null	24	data/indoor01/GT/images
indoor02	Exp1_10_ISO3200	low_light	10	3,200	25	data/indoor02/Exp1_10_ISO3200/images
indoor02	Exp1_20_ISO3200	low_light	20	3,200	25	data/indoor02/Exp1_20_ISO3200/images
indoor02	Exp1_30_ISO3200	low_light	30	3,200	25	data/indoor02/Exp1_30_ISO3200/images
indoor02	GT	reference	null	null	25	data/indoor02/GT/images
leaf	Exp1_100_ISO3200	low_light	100	3,200	24	data/leaf/Exp1_100_ISO3200/images
leaf	Exp1_25_ISO3200	low_light	25	3,200	24	data/leaf/Exp1_25_ISO3200/images
leaf	Exp1_40_ISO3200	low_light	40	3,200	24	data/leaf/Exp1_40_ISO3200/images
leaf	GT	reference	null	null	24	data/leaf/GT/images
pavilion	Exp1_1000_ISO3200	low_light	1,000	3,200	24	data/pavilion/Exp1_1000_ISO3200/images
pavilion	Exp1_200_ISO3200	low_light	200	3,200	24	data/pavilion/Exp1_200_ISO3200/images
pavilion	Exp1_400_ISO3200	low_light	400	3,200	24	data/pavilion/Exp1_400_ISO3200/images
pavilion	GT	reference	null	null	24	data/pavilion/GT/images
stair	Exp1_160_ISO3200	low_light	160	3,200	25	data/stair/Exp1_160_ISO3200/images
stair	Exp1_250_ISO3200	low_light	250	3,200	25	data/stair/Exp1_250_ISO3200/images
stair	Exp1_500_ISO3200	low_light	500	3,200	25	data/stair/Exp1_500_ISO3200/images
stair	GT	reference	null	null	25	data/stair/GT/images
stone	Exp1_100_ISO3200	low_light	100	3,200	25	data/stone/Exp1_100_ISO3200/images
stone	Exp1_40_ISO3200	low_light	40	3,200	25	data/stone/Exp1_40_ISO3200/images
stone	Exp1_60_ISO3200	low_light	60	3,200	25	data/stone/Exp1_60_ISO3200/images
stone	GT	reference	null	null	25	data/stone/GT/images
table	Exp1_1250_ISO3200	low_light	1,250	3,200	23	data/table/Exp1_1250_ISO3200/images
table	Exp1_800_ISO3200	low_light	800	3,200	23	data/table/Exp1_800_ISO3200/images
table	GT	reference	null	null	23	data/table/GT/images
01	Exp1_10_ISO3200	low_light	10	3,200	null	null
01	Exp1_10_ISO3200	low_light	10	3,200	null	null
01	Exp1_10_ISO3200	low_light	10	3,200	null	null
01	Exp1_10_ISO3200	low_light	10	3,200	null	null
01	Exp1_10_ISO3200	low_light	10	3,200	null	null
01	Exp1_10_ISO3200	low_light	10	3,200	null	null
01	Exp1_10_ISO3200	low_light	10	3,200	null	null
01	Exp1_10_ISO3200	low_light	10	3,200	null	null
01	Exp1_10_ISO3200	low_light	10	3,200	null	null
01	Exp1_10_ISO3200	low_light	10	3,200	null	null
01	Exp1_10_ISO3200	low_light	10	3,200	null	null
01	Exp1_10_ISO3200	low_light	10	3,200	null	null
01	Exp1_10_ISO3200	low_light	10	3,200	null	null
01	Exp1_10_ISO3200	low_light	10	3,200	null	null
01	Exp1_10_ISO3200	low_light	10	3,200	null	null
01	Exp1_10_ISO3200	low_light	10	3,200	null	null
01	Exp1_10_ISO3200	low_light	10	3,200	null	null
01	Exp1_10_ISO3200	low_light	10	3,200	null	null
01	Exp1_10_ISO3200	low_light	10	3,200	null	null
01	Exp1_10_ISO3200	low_light	10	3,200	null	null
01	Exp1_10_ISO3200	low_light	10	3,200	null	null
01	Exp1_10_ISO3200	low_light	10	3,200	null	null
01	Exp1_10_ISO3200	low_light	10	3,200	null	null
01	Exp1_10_ISO3200	low_light	10	3,200	null	null
01	Exp1_10_ISO3200	low_light	10	3,200	null	null
01	Exp1_10_ISO3200	low_light	10	3,200	null	null
01	Exp1_10_ISO3200	low_light	10	3,200	null	null
01	Exp1_25_ISO3200	low_light	25	3,200	null	null
01	Exp1_25_ISO3200	low_light	25	3,200	null	null
01	Exp1_25_ISO3200	low_light	25	3,200	null	null
01	Exp1_25_ISO3200	low_light	25	3,200	null	null
01	Exp1_25_ISO3200	low_light	25	3,200	null	null
01	Exp1_25_ISO3200	low_light	25	3,200	null	null
01	Exp1_25_ISO3200	low_light	25	3,200	null	null
01	Exp1_25_ISO3200	low_light	25	3,200	null	null
01	Exp1_25_ISO3200	low_light	25	3,200	null	null
01	Exp1_25_ISO3200	low_light	25	3,200	null	null
01	Exp1_25_ISO3200	low_light	25	3,200	null	null
01	Exp1_25_ISO3200	low_light	25	3,200	null	null
01	Exp1_25_ISO3200	low_light	25	3,200	null	null
01	Exp1_25_ISO3200	low_light	25	3,200	null	null
01	Exp1_25_ISO3200	low_light	25	3,200	null	null
01	Exp1_25_ISO3200	low_light	25	3,200	null	null
01	Exp1_25_ISO3200	low_light	25	3,200	null	null

End of preview.

YAML Metadata Warning:The task_categories "computer-vision" is not in the official list: text-classification, token-classification, table-question-answering, question-answering, zero-shot-classification, translation, summarization, feature-extraction, text-generation, fill-mask, sentence-similarity, text-to-speech, text-to-audio, automatic-speech-recognition, audio-to-audio, audio-classification, audio-text-to-text, voice-activity-detection, depth-estimation, image-classification, object-detection, image-segmentation, text-to-image, image-to-text, image-to-image, image-to-video, unconditional-image-generation, video-classification, reinforcement-learning, robotics, tabular-classification, tabular-regression, tabular-to-text, table-to-text, multiple-choice, text-ranking, text-retrieval, time-series-forecasting, text-to-video, image-text-to-text, image-text-to-image, image-text-to-video, visual-question-answering, document-question-answering, zero-shot-image-classification, graph-ml, mask-generation, zero-shot-object-detection, text-to-3d, image-to-3d, image-feature-extraction, video-text-to-text, keypoint-detection, visual-document-retrieval, any-to-any, video-to-video, other

MILL3D

Dataset Summary

MILL3D is a real-world multi-exposure benchmark for evaluating low-light novel view synthesis (NVS) and 3D reconstruction from moderate low light to extreme darkness. The primary goal of this dataset is evaluation and diagnosis of robustness, not large-scale model training.

Each scene contains multiple capture levels under different exposure settings, together with a high-quality long-exposure reference level (GT). This organization supports exposure-by-exposure comparison of failure modes under controlled real captures.

Supported Tasks

Low-light novel view synthesis benchmark evaluation
Low-light 3D reconstruction benchmark evaluation
Exposure-level robustness analysis

Dataset Structure

The dataset is organized as:

/{scene_name}/{level_name}/images/{image_name}

Examples:

/03/Exp1_20_ISO3200/images/DSC09722.png
/03/GT/images/DSC09722.png

Where:

scene_name identifies a real capture scene
level_name identifies one exposure/capture setting
GT is the reference long-exposure level
paired images across levels share the same image_name

Metadata Files

This repository includes the following minimal metadata files:

metadata/scenes.csv
metadata/levels.csv
metadata/samples.csv
metadata/dataset_summary.json

metadata/samples.csv is the main file for indexing benchmark samples and contains:

scene_name
level_name
capture_type
exposure_value
iso_value
image_name
relative_image_path
reference_level_name
reference_image_path
split

Benchmark Usage

MILL3D is intended primarily as a benchmark dataset for controlled evaluation. Users should preserve the provided scene / level / image pairing when reporting results.

Data Splits

Current metadata marks all entries as:

split = benchmark

If you later define official evaluation subsets, you can extend metadata/samples.csv with more detailed split labels.

License

This dataset is released under CC-BY-4.0.

Citation

If you use MILL3D, please cite the associated paper.

@misc{mill3d2026,
  title={MILL3D: A Real-World Multi-Exposure Benchmark for Extreme Low-Light Novel View Synthesis},
  author={TBD},
  year={2026}
}

Acknowledgements

This dataset was collected for benchmarking robustness of low-light NVS and 3D reconstruction systems under real extreme low-light conditions.

Downloads last month: 23

Total file size:

198 kB