Hokkaido Agriculture Image Dataset: Grape, Wheat, Onion, Apple and Haskap Field Photos with COCO Detection and Segmentation Labels (Hokkaido University and Sony)
- Provider
- Hokkaido University and Sony Group Corporation
- License
- CC-BY-4.0 (source)
- Size
- 1,883 files
- Format
- PNG, JSON, JPEG (in sampled files)
- Stored at
- Hugging Face (https://huggingface.co/datasets/Sony/Hokkaido_Agriculture_Image_Dataset)
- Access
- Free and public. No account with the data provider is needed.
- Category
- Agriculture & Environment
- Source
- https://huggingface.co/datasets/Sony/Hokkaido_Agriculture_Image_Dataset
About this dataset
Field photographs of five crops grown on farms in Hokkaido, Japan, collected jointly by Hokkaido University (Vehicle Robotics Laboratory) and Sony Group Corporation for agricultural computer vision: crop counting, wheat heading detection and robotic harvesting and pruning. About 1,870 images cover wine grapes (1,061 images, with instance segmentation of grape bunches, branches, wires and cut points where a bunch joins the cane, plus a separate object detection set), wheat ears (255), onions (336), apples (171) and haskap honeysuckle berries and flowers (51). Photos span growth stages, weather, lighting and viewpoints, at resolutions from 640 x 480 up to 4896 x 3672. For grapes, apples and haskap only the foremost plant is annotated, so background fruit acts as negative samples. Each crop folder (grape_instance_segmentation, grape_object_detection, wheat, onion, apple, haskap) holds images/ (JPEG, PNG or BMP) and annotation/annotation_coco.json in COCO format, readable with pycocotools; the repository is about 8.9 GB. Useful for fruit detection, yield estimation, harvest point detection for picking robots and wheat head counting. Released by the authors under CC BY 4.0; credit Hokkaido University and Sony Group Corporation.
How to load this dataset in Python
from huggingface_hub import hf_hub_download
path = hf_hub_download(repo_id="Sony/Hokkaido_Agriculture_Image_Dataset", filename="Dataset_summary.png", repo_type="dataset")
print(path)Data structure
File types in sampled files: PNG, JSON, JPEG
Files
Example file: Dataset_summary.png
Analyze this dataset without downloading it
On Scigantic, Hokkaido Agriculture Image Dataset: Grape, Wheat, Onion, Apple and Haskap Field Photos with COCO Detection and Segmentation Labels (Hokkaido University and Sony) opens in a hosted JupyterLab notebook with the data already mounted, so you can read files where they are stored instead of copying them first. Each notebook includes an AI assistant that can see the mounted files and write and run analysis code. Open the dataset page to launch a notebook, or browse the full catalog.
This page is also available as Markdown and as JSON.
Related Agriculture & Environment datasets
- Forest Data Partnership Commodity Probability Maps: Cocoa, Coffee, Oil Palm, Rubber at 10 m, 2018 and 2024 (model 2026a)
- AgricultureVision
- Croppie Coffee Cherry Detection Dataset: Smartphone Photos of Arabica and Robusta Branches with Ripeness-Class Bounding Boxes, Uganda (Producers Direct and Alliance Bioversity-CIAT)
- Deoxynivalenol (DON) Mycotoxin in Wheat by Near-Infrared (NIR) Spectroscopy: 408 Samples from Austria and France (Food AI Nexus Curation)
- Seed Germination Detection Time Series: 23,797 Petri Dish Images of Maize, Rye and Pearl Millet Seeds with Germinated Bounding Boxes (TU Munich, AgML Mirror)
- RaspberrySet: Raspberry (Rubus idaeus) Bud, Flower and Berry Ripeness Bounding Boxes in 2,039 iPhone Images, Latvia (Institute of Horticulture, AgML Mirror)
- MH-Weed16: Indian Soybean Field Weed Images with 62,052 Bounding Boxes across 16 Weed Species, Maharashtra (COEP Technological University, AgML Mirror)
- Corn Kernel Counting Dataset