Naiscorp/car-damage-dataset

Car Damage Images

A raw image collection for vehicle damage assessment. Unlabeled: these images have no annotations yet and are intended as source material for labeling or pre-training.

Structure

images/
  001/  images_001.jpg  images_002.jpg  ...
  002/  ...
  ...                                        683 folders
thumbnails/
  001/  thumbnail_001.jpg  thumbnail_002.jpg  ...
  ...                                        683 folders

Branch	Files	Folders	Size
images/	22,491	683	3.8 GB
thumbnails/	22,303	683	1.5 GB

Important note

images/ and thumbnails/ are NOT paired 1:1. Folder names are identical in both branches, but the number of files inside differs in 210 / 683 folders (188 files difference in total). In some folders, thumbnails/ actually contains more files than images/. Do not assume that images/NNN/images_K.jpg corresponds to thumbnails/NNN/thumbnail_K.jpg; indices run independently within each branch.

If you need the original images, use images/ and ignore thumbnails/.

Download

from huggingface_hub import snapshot_download

# everything
path = snapshot_download("Naiscorp/car-damage-dataset", repo_type="dataset")

# original images only
path = snapshot_download(
    "Naiscorp/car-damage-dataset",
    repo_type="dataset",
    allow_patterns="images/*",
)

hf download Naiscorp/car-damage-dataset --repo-type dataset --local-dir ./car-damage

Load with datasets as an imagefolder (no labels):

from datasets import load_dataset

ds = load_dataset("imagefolder", data_dir=f"{path}/images", split="train")
ds[0]["image"]   # PIL.Image