Research on postharvest tomato freshness recognition method based on RGB-S and ResNet34.
Yuhua Huang1, Juntao Xiong1, Xinjing Jiang1
1College of Mathematics and Informatics, South China Agricultural University, Guangzhou, China.
Journal of Food Science
|March 7, 2025
Summary
Accurate postharvest tomato freshness identification is now possible using a computer vision method. This nondestructive technique enhances postharvest management for growers and ensures consumers receive high-quality produce.
Area of Science:
- Agricultural Engineering
- Computer Vision
- Food Science
Background:
- Accurate postharvest tomato freshness identification is crucial for efficient storage, transportation, and wholesale.
- Current methods may be destructive or lack real-time capabilities.
Purpose of the Study:
- To develop a nondestructive method for identifying postharvest tomato freshness.
- To improve postharvest management practices and reduce produce spoilage.
Main Methods:
- Utilized improved frequency-tuned (FT) visual saliency detection and a ResNet34 model.
- Extracted L*, Y, and H color components as features.
- Combined saliency maps with RGB image data for a four-channel ResNet model.
Main Results:
- Achieved high accuracy (98.38%), precision (98.69%), and recall (98.32%).
- Demonstrated a rapid detection speed of 0.0326 seconds per image.
- Validated the method's effectiveness and real-time performance.
Conclusions:
- The proposed computer vision method offers an effective and fast solution for assessing postharvest tomato freshness.
- This technology supports the fruit and vegetable industry in quality control and supply chain management.
- Provides consumers with a tool to identify fresh tomatoes during purchase.
