webbrain-one/food-dataset

Food Dataset

An image classification dataset of food photos organized into 201 categories (folders), with 35,046 images total (~924 MB).

Each top-level folder is a category (e.g. adana kebab, sushi, waffles, tiramisu, ...) containing JPEG images of that food/dish. This follows the standard Hugging Face imagefolder layout, so it loads directly with:

from datasets import load_dataset

ds = load_dataset("webbrain-one/food-dataset")

Structure

<category name>/<image>.jpg
...
metadata/            # original auxiliary files from the source collection
  labels.txt          # list of label names from an earlier version of the dataset
  train.txt           # partial filename -> label-index mapping from an earlier version
  label_former.php     # original labeling script
  dir_lister.php        # original directory-listing script

License

Private — all rights reserved. This dataset is not released under an open license. It is shared publicly for visibility, but no reuse, redistribution, or derivative use is granted without the owner's explicit permission.

Notes

• Images per category range from 3 to 781 (median 101) — class distribution is imbalanced.
• The metadata/ folder is kept for reference/provenance and is not part of the image-classification splits.