Olfactory Visualization Sensing Array Made with CelluMOFs to Predict Fruit Ripeness Using Deep Learning

Area of Science:

• Materials Science
• Chemical Sensing
• Artificial Intelligence

Background:

• Developing sensitive and accurate artificial scent systems for fruit ripeness detection is challenging.
• Existing methods often lack the required sensitivity and pattern recognition capabilities for on-site assessment.
• Colorimetry-based systems offer potential but require enhanced sensitivity and integration with advanced algorithms.

Purpose of the Study:

• To construct a flexible dye/CelluMOFs-based sensor array for high-sensitivity detection of fruit volatile compounds.
• To integrate a densely connected convolutional network (DenseNet) for accurate recognition of fruit scent fingerprints.
• To achieve precise categorization of fruit ripeness using an olfactory visual sensing system.

Main Methods:

• Synthesis of CelluMOFs via in situ growth of γ-cyclodextrin metal-organic frameworks (γ-CD-MOFs) on filter paper.
• Fabrication of a flexible, porous dye/CelluMOFs sensitive membrane with enhanced dye loading capacity.
• Integration of a DenseNet model with the colorimetric sensor array for pattern recognition and classification.
• Detection of characteristic fruit odors, including trans-2-hexenal, and assessment of sensor stability.

Main Results:

• The CelluMOFs membrane showed a 62x higher specific surface area and 3x increased dye loading capacity compared to filter paper.
• The sensor array demonstrated high sensitivity with low gas detection thresholds for trans-2-hexenal (8-1500 ppm).
• The integrated system achieved a 99.09% classification accuracy for fruit ripeness prediction on the validation set.

Conclusions:

• A novel, flexible dye/CelluMOFs sensor array coupled with DenseNet enables sensitive and accurate fruit ripeness detection.
• The olfactory visual sensing system effectively recognizes unique scent fingerprints for high-precision fruit categorization.
• This approach offers a promising solution for on-site, non-destructive assessment of fruit quality.