Dataset Viewer
The dataset viewer is not available for this dataset.
Cannot get the config names for the dataset.
Error code:   ConfigNamesError
Exception:    BadZipFile
Message:      zipfiles that span multiple disks are not supported
Traceback:    Traceback (most recent call last):
                File "/src/services/worker/src/worker/job_runners/dataset/config_names.py", line 67, in compute_config_names_response
                  config_names = get_dataset_config_names(
                      path=dataset,
                      token=hf_token,
                  )
                File "/usr/local/lib/python3.14/site-packages/datasets/inspect.py", line 161, in get_dataset_config_names
                  dataset_module = dataset_module_factory(
                      path,
                  ...<4 lines>...
                      **download_kwargs,
                  )
                File "/usr/local/lib/python3.14/site-packages/datasets/load.py", line 1182, in dataset_module_factory
                  raise e1 from None
                File "/usr/local/lib/python3.14/site-packages/datasets/load.py", line 1157, in dataset_module_factory
                  ).get_module()
                    ~~~~~~~~~~^^
                File "/usr/local/lib/python3.14/site-packages/datasets/load.py", line 647, in get_module
                  module_name, default_builder_kwargs = infer_module_for_data_files(
                                                        ~~~~~~~~~~~~~~~~~~~~~~~~~~~^
                      data_files=data_files,
                      ^^^^^^^^^^^^^^^^^^^^^^
                      path=self.name,
                      ^^^^^^^^^^^^^^^
                      download_config=self.download_config,
                      ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
                  )
                  ^
                File "/usr/local/lib/python3.14/site-packages/datasets/load.py", line 302, in infer_module_for_data_files
                  split: infer_module_for_data_files_list(data_files_list, download_config=download_config)
                         ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
                File "/usr/local/lib/python3.14/site-packages/datasets/load.py", line 246, in infer_module_for_data_files_list
                  return infer_module_for_data_files_list_in_archives(data_files_list, download_config=download_config)
                File "/usr/local/lib/python3.14/site-packages/datasets/load.py", line 273, in infer_module_for_data_files_list_in_archives
                  f.split("::")[0] for f in xglob(extracted, recursive=True, download_config=download_config)
                                            ~~~~~^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
                File "/usr/local/lib/python3.14/site-packages/datasets/utils/file_utils.py", line 1051, in xglob
                  fs, *_ = url_to_fs(urlpath, **storage_options)
                           ~~~~~~~~~^^^^^^^^^^^^^^^^^^^^^^^^^^^^
                File "/usr/local/lib/python3.14/site-packages/fsspec/core.py", line 436, in url_to_fs
                  fs = filesystem(protocol, **inkwargs)
                File "/usr/local/lib/python3.14/site-packages/fsspec/registry.py", line 329, in filesystem
                  return cls(**storage_options)
                File "/usr/local/lib/python3.14/site-packages/fsspec/spec.py", line 128, in __call__
                  obj = super().__call__(*args, **kwargs, **strip_tokenize_options)
                File "/usr/local/lib/python3.14/site-packages/fsspec/implementations/zip.py", line 70, in __init__
                  self.zip = zipfile.ZipFile(
                             ~~~~~~~~~~~~~~~^
                      self.fo,
                      ^^^^^^^^
                  ...<3 lines>...
                      compresslevel=compresslevel,
                      ^^^^^^^^^^^^^^^^^^^^^^^^^^^^
                  )
                  ^
                File "/usr/local/lib/python3.14/zipfile/__init__.py", line 1472, in __init__
                  self._RealGetContents()
                  ~~~~~~~~~~~~~~~~~~~~~^^
                File "/usr/local/lib/python3.14/zipfile/__init__.py", line 1535, in _RealGetContents
                  endrec = _EndRecData(fp)
                File "/usr/local/lib/python3.14/zipfile/__init__.py", line 375, in _EndRecData
                  return _EndRecData64(fpin, filesize - sizeEndCentDir, endrec)
                File "/usr/local/lib/python3.14/zipfile/__init__.py", line 303, in _EndRecData64
                  raise BadZipFile("zipfiles that span multiple disks are not supported")
              zipfile.BadZipFile: zipfiles that span multiple disks are not supported

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.

S24 Raw-sRGB Dataset

Community mirror of the dataset introduced in Time-Aware Auto White Balance in Mobile Photography (ICCV 2025) by Mahmoud Afifi, Luxi Zhao, Abhijith Punnappurath, Mohamed Ashraf Abdelsalam, Ran Zhang, and Michael S. Brown, AI Center–Toronto, Samsung Electronics.

This mirror is maintained by Lukhausen. It is not the official author repository, and no author or Samsung endorsement is implied.

Mirror availability

Upload complete. All 21 original archive parts in archive-manifest.json are present, with their remote sizes and SHA-256 hashes verified. See SHA256SUMS.

The archive payloads are redistributed without image processing, recompression, relabeling, or changes to the original splits. This dataset card, archive manifest, upload layout, and checksum file are mirror additions.

Contents

The dataset contains 3,224 scenes captured with the main camera of a Samsung Galaxy S24 Ultra, with noisy and denoised raw RGB images, in-camera sRGB outputs, expert Lightroom renderings, five additional Lightroom picture styles, masks, and metadata including white-balance annotations.

The published raw RGB images are demosaiced, black-level-normalized, 16-bit, three-channel PNG files, rather than original DNG sensor files. sRGB outputs are 8-bit JPEGs. According to the authors, personal information has been blurred in both raw and sRGB images. A slight raw/sRGB pixel shift can occur because of in-camera ISP cropping.

Split Scenes Archive parts Archive bytes
train 2,619 15 308,985,969,260
val 205 2 24,300,049,809
test 400 4 47,851,902,177

The split counts are derived from the original ZIP inventories. In total, the original archives contain 38,688 files in 36 folders. The archives occupy approximately 355 GiB, and extracted contents occupy approximately 358 GiB. Retaining both requires approximately 713 GiB.

Each extracted split contains:

raw_images/           # processed raw RGB, 16-bit PNG
denoised_raw_images/  # Lightroom-denoised raw RGB, 16-bit PNG
srgb_images/          # in-camera Pro-mode JPEGs
srgb_images_style_0/  # manually rendered Lightroom JPEGs
srgb_images_style_1/  # five additional Lightroom styles (1 through 5)
...
srgb_images_style_5/
masks/                # secondary-illuminant regions
blur_masks/           # regions blurred for privacy
data/                 # JSON metadata and annotations

See UPSTREAM_README.md and the official dataset page for the complete field definitions, processing details, and limitations. This mirror retains the upstream annotation text; figure links in that copy point to the original author repository.

Download and extract

Download all files in archives/<split>/ for each required split and keep the parts together. For example, validation requires both val.z01 and val.zip; training requires train.z01 through train.z14 plus train.zip.

from huggingface_hub import snapshot_download

snapshot_download(
    repo_id="Lukhausen/S24-Raw-sRGB",
    repo_type="dataset",
    local_dir="S24-Raw-sRGB",
    allow_patterns=["archives/val/*", "archive-manifest.json", "SHA256SUMS", "LICENSE.md"],
)

Extract each split using a recent version of 7-Zip, starting with its .zip file:

7z x S24-Raw-sRGB/archives/val/val.zip -oS24-Raw-sRGB/val
7z x S24-Raw-sRGB/archives/test/test.zip -oS24-Raw-sRGB/test
7z x S24-Raw-sRGB/archives/train/train.zip -oS24-Raw-sRGB/train

SHA256SUMS lists all 21 archive parts. verification/source-file-inventory.jsonl inventories all 38,688 extracted source paths, original byte lengths and ZIP CRC32 values. Checksums describe the mirrored files; the authors did not supply independent checksums in the linked dataset README. The split ZIP layout is intended for downloading and extraction and does not provide a native Hugging Face dataset viewer.

License and attribution

The upstream repository publishes Creative Commons Attribution–NonCommercial–ShareAlike 4.0 International (CC BY-NC-SA 4.0). The upstream dataset README does not state a separate license. This mirror retains that published license and the original author attribution. See LICENSE.md for the full terms and the license explanation.

Reuse must respect the noncommercial restriction, attribution requirements, and applicable ShareAlike conditions. Public availability does not grant unrestricted commercial-use permission. The license does not provide warranties or automatically clear third-party privacy, publicity, trademark, patent, or other rights.

For additional permissions or questions about the original dataset, contact Mahmoud Afifi (m.afifi1@samsung.com, m.3afifi@gmail.com) or Luxi Zhao (lucy.zhao@samsung.com, lucyzhao.zlx@gmail.com), as listed in the official documentation.

Citation

The authors request citation of their paper when the code or dataset is used in research:

@inproceedings{afifi2025time-aware-awb,
  title={Time-Aware Auto White Balance in Mobile Photography},
  author={Afifi, Mahmoud and Zhao, Luxi and Punnappurath, Abhijith and Abdelsalam, Mohamed Ashraf and Zhang, Ran and Brown, Michael S.},
  booktitle={The IEEE International Conference on Computer Vision (ICCV)},
  year={2025}
}

Release verification

The independent verification report records the audited revisions, complete file counts, archive/image hashes, preserved annotations and exact score-to-image joins. Every source path was accounted for. All 3,224 resized annotations were downloaded and compared with CRC-checked originals, and all measurement tables were downloaded and checked. Image/archive hashes were matched against the Hub's content-addressed LFS metadata; anonymous download checks cover all 21 archive parts and image headers from every image/mask folder in every split. The report documents the method; it does not claim a second full download of all image/archive bytes.

Downloads last month
285

Paper for Lukhausen/S24-Raw-sRGB