---
dataset_info:
  features:
  - name: image
    dtype: image
  - name: objects
    struct:
    - name: bbox
      list:
        list: float64
    - name: categories
      list:
        class_label:
          names:
            '0': tomato
  splits:
  - name: train
    num_bytes: 188655796
    num_examples: 895
  download_size: 188661330
  dataset_size: 188655796
configs:
- config_name: default
  data_files:
  - split: train
    path: data/train-*
license: cc0-1.0
task_categories:
- object-detection
size_categories:
- n<1K
---
# Tomato Detection Kaggle

Tomato fruit detection: 895 greenhouse photos of tomato plants (500 x 400 or 400 x 500 pixels) with a bounding box around every visible fruit, 4,930 boxes in a single 'tomato' class (1 to 80 per image), covering ripe and unripe fruit under natural light with leaves, stems and shading as clutter. The images were published on Kaggle by andrewmvd; the MDPI Agronomy study that used them describes them as field photographs taken in a glass greenhouse at the National Engineering Research Center for Facility Agriculture on Chongming Island, Shanghai, China. The dataset contains 895 images with 4,930 bounding box annotations across 1 category.

This dataset is indexed on https://project-agml.github.io/ as part of the AgML python library.

## Citation

```bibtex
@article{wang2025swmd,
  title={SWMD-YOLO: A Lightweight Model for Tomato Detection in Greenhouse Environments},
  author={Wang, Quan and Hua, Ye and Lou, Qiongdan and Kan, Xi},
  journal={Agronomy},
  volume={15},
  pages={1593},
  year={2025},
  publisher={MDPI}
}
```


*This dataset was reformatted from its original format to match HuggingFace standards.*