Utilizing Deep Learning Algorithms to Analyze and Detect Tomato Features in Horticulture.

Abstract

Tomatoes fruits, a pivotal constituent of tomato plants, with a primary emphasis on elucidating the mechanisms governing their quality formation during the ripening process. Against the backdrop of heightened interest in the tomato industry, the research endeavours to augment the efficacy and success of automatic detection under greenhouse tomato conditions, a pivotal facet for the progression of contemporary agricultural practices. The paper introduces a ground-breaking method using convolutional neural networks to accurately classify tomato fruits based on their ripeness and overall condition. The study adopts a modified ResNet50V2 architecture as the underpinning framework for the CNN model, renowned for its effectiveness in image classification tasks. The outcomes demonstrate a commendable 95.36% accuracy in categorizing tomato fruits into four distinct classes: unripe, ripe, old, and damaged.