Dataset · ethz
Food-101: 101 food categories
ethz/food101
101,000 food images across 101 categories for training and evaluating classifiers.
What's inside
The collection contains 101 food categories and 101,000 images, with 750 training and 250 manually verified test images per class. Training images were intentionally left uncleaned, featuring vibrant colors and occasional incorrect labels. All images are scaled so the maximum side is no more than 512 pixels.
What you can do with it
- Train or fine-tune CNNs and ViTs for 101-class food classification.
- Evaluate model robustness against noisy labels in train.
- Compare image classification models on the Food-101 test split.
- Tasks
- Image classification
- Languages
- English
- Size
- 100K–1M rows
- Downloads this month
- 35K
- Likes
- 160
Models trained on it
- tanganke/clip-vit-base-patch32_food101 386
- tanganke/clip-vit-base-patch16_food101 220
- prithivMLmods/Food-101-93MImage classification 171
- tanganke/clip-vit-large-patch14_food101 152
- juan1211/pratos-convnext-tiny-food101Image classification 141
- Lumia101/Food101-EfficientNet-B0Image classification 113
- wongtk/savor-image-generation-modelText → image 53
- Driw0x/my_awesome_food_modelImage classification 52
How to load it
from datasets import load_dataset
ds = load_dataset("ethz/food101")