Articles
A deep learning-based web application to quantify blueberry internal bruising
Article number
1360_26
Pages
211 – 218
Language
English
Abstract
Today, more than ever, blueberries produced in the United States are being machine-harvested for fresh market.
Compared to the hand harvested blueberries, those blueberries harvested mechanically have more internal bruise damage and shorter shelf life.
Changes in harvester design, particularly in fruit catching system and surface, have reduced internal bruise damage.
However, hand harvesting still results in superior quality blueberries.
Researchers in the US have a better understanding of what causes bruise damage during harvesting.
Some are working with harvesting and packing equipment manufacturer to reduce bruise damage in blueberries.
In this study, we developed a deep learning-based web application that can automatically determine the blueberry bruise levels so the users can evaluate the mechanical harvesters in the field as well as packing lines.
We annotated training images to train convolutional neural network models that were used to detect and segment berries and bruised areas from the cross-section of the sliced berries.
The average precision (AP) for the berry detection model was 0.977, while the mean intersection over union (IoU) for berry segmentation and bruise segmentation was 0.979 and 0.773, respectively, for the validation data set.
The linear regression showed a high correlation between the bruise ratio predicted by the machine learning model and the ground truth that was manually annotated.
The developed deep-learning based web application is a robust tool for blueberry breeders and growers to evaluate berry internal bruises created by mechanical harvesters in the field.
Compared to the hand harvested blueberries, those blueberries harvested mechanically have more internal bruise damage and shorter shelf life.
Changes in harvester design, particularly in fruit catching system and surface, have reduced internal bruise damage.
However, hand harvesting still results in superior quality blueberries.
Researchers in the US have a better understanding of what causes bruise damage during harvesting.
Some are working with harvesting and packing equipment manufacturer to reduce bruise damage in blueberries.
In this study, we developed a deep learning-based web application that can automatically determine the blueberry bruise levels so the users can evaluate the mechanical harvesters in the field as well as packing lines.
We annotated training images to train convolutional neural network models that were used to detect and segment berries and bruised areas from the cross-section of the sliced berries.
The average precision (AP) for the berry detection model was 0.977, while the mean intersection over union (IoU) for berry segmentation and bruise segmentation was 0.979 and 0.773, respectively, for the validation data set.
The linear regression showed a high correlation between the bruise ratio predicted by the machine learning model and the ground truth that was manually annotated.
The developed deep-learning based web application is a robust tool for blueberry breeders and growers to evaluate berry internal bruises created by mechanical harvesters in the field.
Authors
X. Ni, F. Takeda, H. Jiang, W.Q. Yang, S. Saito, C. Li
Keywords
blueberry bruise, web application, deep learning, machine learning, computer vision
Groups involved
- Division Precision Horticulture and Engineering
- Division Plant-Environment Interactions in Field Systems
- Division Vine and Berry Fruits
- Division Tropical and Subtropical Fruit and Nuts
- Division Plant Genetic Resources, Breeding and Biotechnology
- Division Temperate Tree Nuts
- Division Temperate Tree Fruits
- Division Vegetables, Roots and Tubers
- Division Horticulture for Development
- Working Group Mechanization, Digitization, Sensing and Robotics
Online Articles (45)