Related Experiment Video
Updated: Jul 29, 2025
11:37
RGB and Spectral Root Imaging for Plant Phenotyping and Physiological Research: Experimental Setup and Imaging Protocols
Published on: August 8, 2017
16.3K
Feasibility Study of Combining Hyperspectral Imaging with Deep Learning for Chestnut-Quality Detection
Qiongda Zhong1,2,3, Hu Zhang1,2,3, Shuqi Tang1,3
1College of Engineering, South China Agricultural University, Guangzhou 510642, China.
Foods (Basel, Switzerland)
|May 27, 2023
Summary
This study introduces a fast method for detecting chestnut quality using hyperspectral imaging (HSI) and deep learning. The FD-UVE-CNN model achieved high accuracy, significantly reducing detection time.
Area of Science:
- Agricultural Science
- Food Science
- Spectroscopy
Background:
- Chestnut quality detection is crucial for processing.
- Traditional methods struggle with quality assessment due to lack of visible symptoms.
Purpose of the Study:
- Develop a rapid, efficient method for chestnut quality detection.
- Utilize hyperspectral imaging (HSI) and deep learning for qualitative and quantitative analysis.
Main Methods:
- Applied principal component analysis (PCA) for qualitative analysis.
- Pre-processed spectral data and constructed traditional machine learning and deep learning models.
- Identified key wavelengths (around 1000, 1400, 1600 nm) for improved efficiency.
Main Results:
- Deep learning models outperformed traditional models, with FD-LSTM reaching 99.72% accuracy.
- The FD-UVE-CNN model achieved 97.33% accuracy after incorporating important wavelengths.
- Using key wavelengths reduced recognition time by an average of 39 seconds.
Conclusions:
- Deep learning combined with HSI shows significant potential for effective chestnut quality detection.
- The FD-UVE-CNN model is identified as the most effective for this application.
