Smartphone AI estimates avocado ripeness with high accuracy
With over 1,400 iPhone images as its dataset, the AI model can help consumers and retailers avoid cutting into overripe avocados.
Researchers at Oregon State University and Florida State University have unveiled a smartphone-based AI system that accurately predicts the ripeness and internal quality of avocados.
They trained models using more than 1,400 iPhone images of Hass avocados, achieving around 92% accuracy for firmness (a proxy for ripeness) and over 84% accuracy in distinguishing fresh from rotten fruit.
Avocado waste is a major issue because they spoil quickly, and many are discarded before reaching consumers. The AI tool is intended to guide both shoppers and businesses on when fruit is best consumed or sold.
Beyond consumer use, the system could be deployed in processing and retail facilities to sort avocados more precisely. For example, more ripe batches might be sent to nearby stores instead of longer transit routes.
The researchers used deep learning (rather than older, manual feature extraction) to capture shape, texture and spatial cues better. As the model dataset grows, its performance is expected to improve further.
Would you like to learn more about AI, tech and digital diplomacy? If so, ask our Diplo chatbot!
