An efficient method for chili pepper variety classification and origin tracing based on an electronic nose and deep...

Area of Science:

• Agricultural Science
• Food Science
• Sensor Technology

Background:

• Chili pepper quality is linked to variety and origin, but market substitution and cross-contamination pose challenges.
• Existing methods for chili pepper identification are often costly, destructive, or require specialized expertise.
• Accurate classification and origin tracing are crucial for ensuring chili pepper quality and authenticity.

Purpose of the Study:

• To develop a novel, accurate, and efficient method for chili pepper variety classification and origin traceability.
• To integrate an electronic nose (e-nose) with a sensor-aware convolutional network (SACNet) for enhanced identification capabilities.
• To overcome the limitations of existing costly and destructive analytical techniques.

Main Methods:

• Utilized an electronic nose (e-nose) to collect and analyze gas samples from various chili peppers.
• Developed a sensor-aware convolutional network (SACNet) incorporating a sensor attention module.
• Implemented a local and wide-area sensing structure within SACNet to capture detailed gas features.

Main Results:

• SACNet achieved high accuracy in variety classification (98.56% on Dataset A) and origin traceability (97.43% on Dataset B, 99.31% on Dataset C).
• The sensor attention module adaptively focused on important sensor data for improved gas information gathering.
• SACNet demonstrated superior performance and efficiency compared to other networks in comparative experiments.

Conclusions:

• The combination of SACNet and e-nose provides an effective and non-destructive strategy for chili pepper identification.
• This technology offers a promising solution for ensuring the quality and authenticity of chili peppers in the market.
• The developed method presents significant advantages in terms of accuracy and parameter efficiency for agricultural product authentication.