Fuzzy logic beats standard AI at spotting new tomato diseases
Rajalakshmi D, Kannan K, Menaga A, Subashini G, Punitha S
Crop Improvement
The tomato plants wilting or spotting in your backyard garden could someday get diagnosed by an app that admits uncertainty instead of guessing wrong with false confidence.
Scientists trained a computer program to recognize sick tomato leaves by looking at color and texture patterns, then comparing them to typical disease examples using a math technique that embraces uncertainty rather than forcing a rigid yes-or-no answer. On familiar test photos, older methods like Random Forest and Support Vector Machines actually scored higher accuracy. But when researchers tested all the models on a completely different set of leaf photos, the fuzzy logic system held up much better, correctly identifying diseases 67% of the time versus around 51-55% for the other methods.
Key Findings
Improved Fuzzy Soft model hit 88.57% accuracy on the original PlantVillage dataset, trailing SVM (97.94%), Random Forest (97.78%), and LDA (94.92%)
On cross-dataset validation with unseen images, the Fuzzy Soft model reached 67.35% accuracy, outperforming SVM (55.10%), LDA (51.02%), and Random Forest (51.02%)
Wilcoxon Signed-Rank testing confirmed the performance gap between the Fuzzy Soft model and Random Forest is statistically significant
chevron_right Technical Summary
Researchers built an AI system that uses fuzzy logic instead of typical deep learning to identify tomato leaf diseases, and while it scored lower on familiar images than standard classifiers, it proved more reliable when tested on unfamiliar photos it hadn't seen before.
Abstract Preview
Original paper
Advancing sustainable agriculture through multi-parameter fuzzy soft set-based plant disease classification.
Plant diseases significantly affect agricultural productivity and global food security, while accurate disease identification remains challenging because of uncertain and overlapping visual symptom...
open_in_new Read full abstractAbstract copyright held by the original publisher.
Species Mentioned
Was this useful?
Want to tell us more? (optional)
Thanks for the note!
Something went wrong — please try again.
Too many submissions. Try again in an hour.
Gene editing removes 97% of celiac-triggering proteins from bread wheat
It could mean that people with celiac disease — roughly 1 in 100 worldwide — may one day safely eat bread made from real wheat, without sacrificing the taste...
The tomato is a plant whose fruit is an edible berry that is eaten as a vegetable. The tomato is a member of the nightshade family that includes tobacco, potato, and chili peppers. It originated from western South America, and may have been domesticated there, in Mexico, or in Central America. Th...