Olfactory Visualization\nSensing Array Made with CelluMOFs\nto Predict Fruit Ripeness Using Deep Learning

Developing a colorimetry-based artificial scent screening\nsystem\n(i.e., an olfactory visual sensing system) with high sensitivity and\naccurate pattern recognition for detecting fruit ripeness remains\nchallenging. In this work, we construct a flexible dye/CelluMOFs-based\nsensor array with improved sensitivity for on-site detection of characteristic\ngases of fruits and integrate a densely connected convolutional network\n(DenseNet) into the sensor array, enabling it to recognize unique\nscent fingerprints and categorize the ripeness of fruits. In the system,\nCelluMOFs are synthesized through in situ growth of γ-cyclodextrin\nmetal–organic frameworks (γ-CD-MOFs) on flexible fiber\nfilter paper to fabricate a uniform, flexible and porous dye/CelluMOFs\nsensitive membrane. Compared to the pristine filter paper, the CelluMOFs\nexhibit increased porosity with a 62 times higher specific surface\narea and a 3-fold increase in dye loading capacity after 12 h of adsorption.\nThe prepared dye/CelluMOFs sensing film shows outstanding mechanical\nand detection stability with negligible deviation after 100 cycles\nof rubbing. The colorimetric visualization arrays with multiple colorimetric\ndye/CelluMOFs chips, enable the sensitive recognition and detection\nof nine kinds of characteristic fruit odors and achieve a high response\nat 8–1500 ppm of <i>trans</i>-2-hexenal, showcasing\nremarkably low gas detection thresholds. On the basis of the ppm-level\nlimit of detection with high sensitivity, the fabricated colorimetric\nsensor arrays are typically used for in situ assessment of fruit ripeness\nby integrating DenseNet. This approach achieves a satisfactory classification\naccuracy of 99.09% on the validation set, enabling high-precision\nprediction of fruit ripeness levels.