[frankinstien]

Here's a paper describing the use of a colorimetric barcode that is made up of 20 different types of porous nanocomposite comprised of dye-loaded chitosan nanoparticles (CNP-dye) embedded on cellulose acetate (CA). The halochromic dyes in the barcode cross-react according to the type and concentration of gases to form a scent fingerprint (in the form of a colorful barcode) that is identified using DCNN (Deep Convolutional Neural Networks). Using fully supervised training techniques and 3475 labeled barcode images as the training source, they obtained a DCNN that predicted meat freshness with an overall accuracy of 98.5%.