CABI Published Agricultural Datasets on Hugging Face

The release aims to improve AI advisory tool accuracy for farmers using region-specific crop pest and disease data.

Updated on Sept. 29, 2026 in Artificial Intelligence

Bold flat-color editorial illustration of stacked botanical leaf samples in laboratory trays, representing structured agricultural data analysis.
The Centre for Agriculture and Biosciences International has released its first AI-ready agricultural datasets to improve diagnostic accuracy for farmers in Kenya, Ethiopia, and India. AI Illustration. Upload story photo >

Live Poll

Do you trust artificial intelligence to provide reliable advice for your daily tasks and decisions?

The Centre for Agriculture and Biosciences International has published its first artificial intelligence-ready datasets on the Hugging Face platform. Developed under the Generative AI for Agriculture project, these resources include information on crop pests and diseases from Kenya, Ethiopia, and India.

Why it matters

Access to verified, locally relevant data is essential for developing AI-driven digital advisory tools that effectively identify crop health issues. This release addresses the critical need for structured agricultural knowledge to improve the performance of automated diagnostic applications.

The released dataset provides structured information on plant health across three countries. It follows a 2025 study of 120 farmers in Kiambu, Kakamega, Meru, and Nakuru, which assessed the utility of AI applications in identifying pests, diseases, and nutrient deficiencies from photographs.

The players

Centre for Agriculture and Biosciences International

A global research organization focused on agricultural development and the dissemination of scientific information.

Hugging Face

A prominent platform for hosting machine learning models, datasets, and collaborative artificial intelligence development tools.

The details

The datasets function by providing AI-ready information that enables machine learning models—computer algorithms that improve through exposure to data—to classify crop threats from user-provided photographs. By training on images and data from Kenya, Ethiopia, and India, these models can identify plant-specific symptoms of disease and malnutrition. This information is intended to support digital advisory tools that offer farmers actionable, location-specific crop health guidance.

Timeline

  1. Kenya launched its Artificial Intelligence Strategy 2025-30 in March 2025.

  2. A study was conducted with 120 farmers across four counties during 2025.

  3. The strategy framework is set to span the 2025-2030 period.

The Tech Race

This release aligns with Kenya's Artificial Intelligence Strategy 2025-30, which prioritizes the digitalization of national economic pillars like agriculture. It marks a push to standardize training data in a sector where locally verified information is currently a primary competitive bottleneck.

These datasets provide developers of agricultural applications with the base information required to build better diagnostic tools for farmers. Users in the agricultural tech space can access the material now to begin testing or integrating the data into existing advisory workflows.

The takeaway

The availability of curated, regional data is a foundational step in moving agricultural AI from research prototypes to field-ready tools. Monitor the project's Hugging Face repository for upcoming releases in poultry and livestock management to gauge the expansion of the platform's utility.

What happens next

The Generative AI for Agriculture project has announced plans to release additional datasets covering poultry and cattle, alongside an expansion of its existing plant-health collection.

Further reading

Explore more developments in data infrastructure and machine learning models within our Artificial Intelligence section.

Source note: This article includes information reported by The Star.

Live Poll

Do you trust artificial intelligence to provide reliable advice for your daily tasks and decisions?