Related Experiment Video
Updated: Jul 2, 2025
08:02
Author Spotlight: Innovative Ice Cream Melting Behavior Analysis Through a Computer Vision System
Published on: October 4, 2024
2.4K
Design and Experimentation of a Machine Vision-Based Cucumber Quality Grader
Fanghong Liu1, Yanqi Zhang2, Chengtao Du1
1State Key Laboratory of Robotics and System, Harbin Institute of Technology, Harbin 150001, China.
Foods (Basel, Switzerland)
|February 24, 2024
Summary
A new machine vision and deep learning cucumber grader protects delicate cucumbers and automates grading. The system uses a novel MassNet model for accurate mass prediction, achieving 93% grading efficiency.
Area of Science:
- Agricultural Engineering
- Computer Vision
- Machine Learning
Background:
- North China type cucumbers are prone to damage during grading due to dense spines and top flowers.
- Manual grading is inefficient, time-consuming, and labor-intensive, impacting market value.
Purpose of the Study:
- To develop an automated cucumber quality grader using machine vision and deep learning.
- To design a novel grading mechanism minimizing damage to cucumbers.
- To introduce a deep learning model for accurate cucumber mass prediction.
Main Methods:
- Designed a novel fixed tray type grading mechanism to prevent damage.
- Developed a convolutional neural network, MassNet, for predicting cucumber mass from top-view images.
- Integrated the mechanism and MassNet for automated mass grading.
Main Results:
- The grader achieved a maximum capacity of 2.3 tons per hour.
- MassNet outperformed AlexNet, MobileNet, and ResNet in mass prediction (MAPE 3.9%, RMSE 6.7 g).
- Online mass grading efficiency reached 93%.
Conclusions:
- The proposed cucumber quality grader effectively automates grading while minimizing damage.
- MassNet offers a robust solution for non-destructive cucumber mass estimation.
- The integrated system significantly improves grading efficiency and preserves cucumber quality.
