Plant Diseases Training Dataset

A large corpus of images of plant leaves with and without diseases, compiled from several open agricultural sources. Perfect for training models for the classification or detection of diseases in agriculture.

Download dataset

Size

116,000 JPEG images, classified by plant type and disease

Licence

CC0: Public Domain

Description

‍

The dataset Plant Diseases Training Dataset brings together more than 116,000 images of plant leaves from several agricultural sources. Each sub-dataset targets a specific crop (potato, rice, cassava, apple, vine...) and offers images annotated according to the pathology visible on the leaf. It is an ideal set for computer vision projects in the agricultural sector.

‍

What is this dataset for?

‍

Develop AI models that can automatically detect plant diseases from images
Improving phytosanitary management on farms
Serve as the basis for a mobile diagnostic aid application

‍

Can it be enriched or improved?

‍

Yes. It is possible to add metadata (location, type of culture, severity level), to annotate the exact contours of the lesions (segmentation) or to increase the data by synthesis (data augmentation). We can also cross-reference this data with weather sensors for richer predictive models.

‍

🔎 In summary

Criterion	Evaluation
🧩 Ease of use	⭐⭐⭐⭐✩ (Well structured by folder and class name)
🧼 Need for cleaning	⭐⭐⭐⭐⭐ (Light: standardize naming formats if needed)
🏷️ Annotation richness	⭐⭐⭐✩✩ (Medium: classification by disease only, without marked zones)
📜 Commercial license	✅ Yes (CC0)
👨‍💻 Beginner friendly	✅ Very good starting point for agricultural computer vision
🔁 Fine-tuning ready	✅ Excellent base for fine-tuning a CNN model
🌍 Cultural diversity	🌱 Good: leaves from multiple regions and crops