# Development of a fruit size estimation method using Mask RCNN for water stress estimation in Satsuma mandarin trees

## TL;DR

This paper introduces a smartphone-based method using Mask RCNN to estimate fruit size in Satsuma mandarin trees, helping manage water stress and improve irrigation.

## Contribution

A novel smartphone-based fruit size estimation method using Mask RCNN and regression for water stress assessment in citrus trees.

## Key findings

- The method achieved a coefficient of determination of 0.988 for estimating fruit diameters.
- A strong correlation (0.7789) was found between fruit growth and water stress.
- The method can integrate with irrigation systems for optimized water use.

## Abstract

Satsuma mandarins can produce high-sugar fruits through effective water stress management during the summer and fall seasons. Estimating water stress is crucial as it allows for precise irrigation management, directly impacting fruit quality and yield. In recent years, the development of automatic irrigation technology has increased the demand for methods to obtain digital data on water stress and related data. This study presents a smartphone-based fruit size estimation method that leverages the correlation between fruit growth and tree water stress. The developed method combined Mask RCNN and a regression model. The coefficients of determination between actual and estimated fruit diameters, between fruit growth and water stress, and between measured diameter and water stress were 0.988, 0.7789, and 0.7349, respectively. This study’s method of accurately estimating fruit diameter using smartphone technology can be integrated with irrigation systems to optimize water usage based on real-time assessments of fruit development and water stress.