Fyffes: Pineapple Sizing for Yield Forecasting

Industry: Agriculture & Supply chain
Solution: Computer Vision for Crop Sizing
Impact: 90% accuracy in pineapple size estimation, enabling 33% improvement in yield forecasting

Overview
Fyffes, one of the world’s leading fresh produce companies, wanted a smarter way to predict pineapple sizes straight from the field—without relying on manual measurement. With a global footprint and a reputation for innovation, they turned to Ishango to help enhance forecasting accuracy with AI.

The Challenge
Accurately estimating pineapple size at scale is a key part of planning for packaging, logistics, and distribution. Fyffes needed a scalable solution that could be used in the field using standard smartphones to capture crop images.

The Ishango Approach
Our team developed a deep learning model using Python that applied transfer learning techniques to recognise and estimate pineapple size from phone-generated images. Hundreds of field images were annotated and used to train the model.

The Result
The final model achieved up to 90% accuracy in estimating fruit size and flagged potential sources of error for further optimisation. This helped Fyffes improve operational planning, reduce waste, and automate a previously manual process.