Computer Vision
Nigerian Agriculture Image Dataset
Crop classification target: maize, cassava, cocoa, tomato, yam and plantain.
Dataset description
Catalogue card for 25,000 field images. This preview ships stylized sensor plates inside the task runner, not photographs. Object storage for images is not connected.
- Creator
- AI HUB Data Team (demo)
- Records
- 25,000
- Format
- JPG + COCO JSON (not hosted)
- Collection period
- Not collected
- Updated
- 2026-11-07
- Downloads of samples
- 0
Quality score 89/100
₦320,000 licence (demo price)
At 20% platform fee: creator ₦256,000, platform ₦64,000. This is a quote. Paystack is not connected.
Samples use a 10-minute grant. Full files are not hosted.
Quality breakdown
20% completeness + 18% accuracy + 14% consistency + 12% annotation agreement + 14% documentation + 12% duplicate health + 10% metadata.
Completeness
86%
Accuracy
91%
Consistency
89%
Duplicate rate
2.4%
Annotation agreement
88%
Documentation
85%
Metadata
83%
Data preview
Safe sample only. Restricted fields are not shown.
| image_id | crop | state | season |
|---|---|---|---|
| plate-maize | maize | Kaduna | rain |
| plate-cassava | cassava | Ogun | rain |
Data card
- Purpose
- Crop classification and bounding-box detection.
- Collection method
- Planned field collection with consent for identifiable farms withheld.
- Annotation method
- Class label plus one box per primary crop.
- Source
- Demo card.
- Size
- 25,000 catalogue target. Sample only is hosted.
- Fields
- image_id, crop, bbox, state, season
- Language
- English labels
- Geography
- Nigeria
- Limitations
- No image files. Plates in tasks are geometric stand-ins.
- Potential bias
- Would under-represent northern dry-season scenes if only southern farms were shot.
- Recommended uses
- Designing an annotation workflow.
- Prohibited uses
- Training a vision model on this card.
- Licence
- commercial
Versions
v0.9 · 2026-03-01
Schema and class list only.
+0 / −0 · Files absent · crop, bbox, state
Report dataset
Use this for privacy, copyright, harmful content or a misleading card. Do not attach more personal data than the report needs.
You can also use the report desk.