1：Experimental Design and Method Selection:

The study utilized visible and near-infrared hyperspectral imaging technology to identify the species and sex of silkworm pupae. Successive projection algorithm (SPA) was used for variable selection, and principal component analysis (PCA) was used to choose characteristic images. Gray-level co-occurrence matrix (GLCM) was implemented on the first three principal component images to extract textural features. Partial least squares discriminant analysis (PLS-DA) and support vector machine (SVM) models were built based on spectral data, textural data, and fusion data.

2：Sample Selection and Data Sources:

288 hyperspectral images of silkworm pupae were collected, including 48 female 9311 × 991, 48 male 9311 × 991, 48 female 9312 × 827, 48 male 9312 × 827, 48 female 9312 × shanhe B, and 48 male 9312 × shanhe B. Samples were divided into a calibration set and a prediction set using the Kennard-Stone (KS) algorithm.

3：List of Experimental Equipment and Materials:

The Vis/NIR HSI system (363–1026 nm) was utilized, consisting of a spectrograph (Imspector V10E, Spectral Imaging Ltd., Oulu, Finland), a high performance Electron-Multiplying Charge Coupled Device (EMCCD, Raptor EM285CL, China), two fiber halogen lamps (IT 3900, 150W), a moving platform driven by a stepping motor (Isuzu Optics Corp., Taiwan, China), and a computer with data acquisition and preprocessing software (Spectral Image software, Isuzu Optics Corp., Taiwan, China).

4：Experimental Procedures and Operational Workflow:

Hyperspectral images were acquired in reflectance mode, corrected, and processed. The average spectra were extracted from the region of interest (ROI) around the tail region of silkworm pupae. Spectral and textural data were extracted, fused, and normalized before modeling.

5：Data Analysis Methods:

PLS-DA and SVM models were built based on spectral data, textural data, and fusion data. Model performance was evaluated by identification accuracy.