This technique implements near-infrared spectroscopy (NIR) and hyperspectral imaging (HSI) with partial least squares discriminant analysis (PLS-DA) and one-dimensional convolutional neural networks (1D-CNN) to establish a non-destructive fish quality assessment model. Through measurements taken from the outer surface with fish scales, the model can identify abnormal flesh conditions affected by Streptococcus, and estimate the affected regions. With further improvements in model robustness and equipment affordability, this technology could provide a practical alternative to conventional invasive inspection, reduce product returns and associated logistics costs, enhance quality control, and support the technological advancement of the aquaculture industry.