I have, in my opinion, a rational and vigorous skepticism about the intrusion of artificial intelligence into our everyday lives. Artificial intelligence, a powerful tool that has the potential to improve the lives of all humanity, seems to be the main goal of artificial intelligence so far in the realm of bad art, hacker tricks and increasing the profitability of companies through reduced labor costs. Those who should be helped by AI are those who have lost their jobs; no one in management positions seems overly concerned that automation will take away from their wages, when in fact there seem to be some real costs that can be reduced by allowing decisions at the management level to be made literally callously.
Still, AI can make things better. In the world of coffee, a promising novel application of artificial intelligence is green coffee detection and defect classification.
In a novel study recently published in the journal researchers sought to test whether YOLO (You Only Look Once), a deep learning model that detects objects in photos or videos, could effectively identify and classify green coffee. To do this, the researchers used various variants of YOLO and trained them on an image bank of over 4,000 images “covering a variety of grain types, defects, and lighting conditions.”
They found that their custom version of the YOLOv8n model – “specifically designed to detect defects in coffee” – performed best on all parameters, with a precision rate of 97.7% (the percentage of “true positives”, i.e. green coffee, detected by the algorithm to all green coffee detections), a recall of 99.9% (the “probability of accurately detecting truth objects”, in this case green coffee), and an f1-score of 98.3% (combination of the first two indicators).
Not only was the algorithm able to determine what green coffee was from the image, but it also accurately identified the four different types of defects — black, cracked, fading and sour — on which it was trained.
YOLO models have previously been used to successfully identify apple blossoms, tomatoes, cherries and apples, but have not yet been applied to coffee production. But the potential benefits are significant. Consistency is the production of coffee and the ability to remove defects are significant factors in increasing the rating of the cup and therefore a higher price for the harvest. The researchers note that emerging manufacturing markets such as Bangladesh could greatly benefit from artificial intelligence, “where it could significantly boost the economy and improve farmers’ living conditions.”
This is perhaps the perfect implementation of artificial intelligence. Performing a time-consuming, almost human-impossible task to add overall value. It doesn’t try to replace the artist, the barista, the producer, or any other very human part of the record. It is a tool thanks to which they can be more effective in their work.
