For TensorFlow Lite model training purposes.
This project does not include a fixed dataset. You can train the model using any image dataset organized in a class-folder structure.
Large-scale dataset containing 101 food categories.
-
Official Source: https://data.vision.ee.ethz.ch/cvl/datasets_extra/food-101/
-
Kaggle Mirror: https://www.kaggle.com/datasets/dansbecker/food-101
Dataset for fruit classification tasks.
- Kaggle Source: https://www.kaggle.com/datasets/moltean/fruits
Japanese food image dataset with bounding box annotations.
- Project Page: http://foodcam.mobi/dataset100.html
You can also use your own dataset for better domain-specific performance.
Example structure:
dataset/images/
├── pizza/
├── burger/
├── pasta/
├── sushi/