“`html
Noninvasive identification of various dried squid attributes through hyperspectral imaging integrated with 1D-KAN-CNN
Summary: Considering that dried squid is a prized seafood item in Asian nations, the international food sector demands a rapid and nonintrusive evaluation of this product’s quality. Thus, this investigation employs visiblenear-infrared (VIS-NIR) hyperspectral imaging alongside deep learning (DL) techniques. We collected and processed VIS-NIR (4001000 nm) hyperspectral reflectance images from 93 dried squid specimens. Significant wavelengths were determined through competitive adaptive reweighted sampling, principal component analysis, and the successive projections method. Utilizing a Kolmogorov-Arnold network (KAN), we present a one-dimensional KAN convolutional neural network (1D-KAN-CNN) for noninvasive analysis of fat, protein, and total volatile basic nitrogen….
“`