Intact Detection of Highly Occluded Immature Tomatoes on Plants Using Deep Learning Techniques.

Yue Mu,Yue Mu,Tai-Shen Chen,Wei Guo,Seishi Ninomiya,Seishi Ninomiya

doi:10.3390/s20102984

Abstract

Automatic detection of intact tomatoes on plants is highly expected for low-cost and optimal management in tomato farming. Mature tomato detection has been wildly studied, while immature tomato detection, especially when occluded with leaves, is difficult to perform using traditional image analysis, which is more important for long-term yield prediction. Therefore, tomato detection that can generalize well in real tomato cultivation scenes and is robust to issues such as fruit occlusion and variable lighting conditions is highly desired. In this study, we build a tomato detection model to automatically detect intact green tomatoes regardless of occlusions or fruit growth stage using deep learning approaches. The tomato detection model used faster region-based convolutional neural network (R-CNN) with Resnet-101 and transfer learned from the Common Objects in Context (COCO) dataset. The detection on test dataset achieved high average precision of 87.83% (intersection over union ≥ 0.5) and showed a high accuracy of tomato counting (R2 = 0.87). In addition, all the detected boxes were merged into one image to compile the tomato location map and estimate their size along one row in the greenhouse. By tomato detection, counting, location and size estimation, this method shows great potential for ripeness and yield prediction.

Highlights

Tomatoes are the second most important horticultural crop [1] in terms of yield, with total production of more than 180 million tonnes across the world (FAO STAT 2017 [2])
Some previous researches precision-recall curve for determining the F1 score [13,14,18,28], but from their accuracy descriptions, lack the F1 score recording or precision-recall curve for determining the F1 score [13,14,18,28], but the accuracy of this research were higher than most machine learning method
The research [19] got a from their accuracy descriptions, the accuracy of this research were higher than most machine high F1 score for mature tomato detection, which was less difficult than for green tomatoes

Summary

Introduction

Tomatoes are the second most important horticultural crop [1] in terms of yield, with total production of more than 180 million tonnes across the world (FAO STAT 2017 [2]). For high yield and good quality, the crop needs precision management of water throughout the growing period [4], as well as fertilizer and pest control [5]. Depending on the ultimate use of the tomatoes, they may be harvested at different stages of ripeness. Optimal tomato cultivation requires tomato-on-plant detection to provide the fruit location and ripening status on spatial variation on which to base agronomic decisions [7]. To inform harvest resourcing and management, and marketing, tomato yield prediction requires dynastically and precise monitoring of tomato number, size, and ripening

Objectives

Methods

Results

Discussion

Conclusion