Detection and Localization of Ripe Tomatoes Using Machine Vision

Abstract

Tomato is one of the most important agricultural products and the most widely used fruit in the world, whose quality for the consumer depends on its appearance. Therefore, timely delivery of tomatoes to the consumer has always been one of the major concerns of farmers. The study tried to design a model for a tomato harvester robot using machine vision to be able to detect and locate tomatoes automatically and in real-time. The algorithm implemented for this purpose is based on the You Only Looked Once Version 4 (YOLO-V4) object detection algorithm. The basic dataset used in the study was 194 images at 416×416 dimensions, increased to 1008 using data augmentation methods and other pre-processing and were used for network training and testing. The use of these methods along with the two methods of Spatial Pyramid Pooling and route aggregation network in this system, despite the small value of the initial training data, has caused the network to reach 86% accuracy according to F1-Score metric for experimental data processing.