Abstract: 【Objective】This study aimed to establish a non-destructive discrimination method for the eggplant seedling wilting degree based on hyperspectral imaging technology，and to conduct research on discrimination of eggplant wilting degree under high temperature stress based on hyperspectral imaging，thereby providing reference for high-throughput precise identification of eggplant heat tolerance and high-throughput screening of heat-tolerant eggplant germplasm resour-ces.【Method】The seedlings of eggplant inbred line KY sensitive to high temperature were used as the research objects. The SOC710VP hyperspectral imaging system（with a spectral resolution of 1.3 nm and a spectral range of 400-1000 nm）was used to collect canopy images of eggplant seedlings at different wilting levels after high temperature stress. The effects of five pretreatment methods，including variable transformation （SNV）， autoscaling （AUT）， normalization （NOR）， Savitzky-Golay smoothing （SG）， and multiplicative scatter correction （MSC）， were compared. Based on the training set and test set divided by the Kennard-Stone algorithm， a wilting level discrimination model was constructed using partial least squares discriminant analysis （PLS-DA） and support vector machine （SVM） with full bands and feature bands extracted by the competitive adaptive reweighted sampling （CARS） algorithm.【Result】Differences were found in the average reflectance of plants at different wilting levels，and the average reflectance in the 400-1000 nm band decreased with the increase of wilting levels. All five pretreatments could effectively reduce the noise of stray light compared with the situation without treatments，among which the SG treatment could remove random signals and improve the signal-to-noise ratio of samples. In the PLS-DA model，the model after SG pretreatment showed the best discrimination effect，with the accuracy of the training data set and test data set reaching 97.00% and 98.00% respectively. The use of CARS to extract feature bands could effectively reduce data dimensions，and the model performance was optimal when combined with SG pretreatment. In the SVM model applying different kernel functions， higher discrimination accuracy was achieved under SNV pretreatment combined with linear kernel function and SG pretreatment combined with polynomial kernel function. Overall，the SG-PLS-DA model had stable performance and high discrimination efficiency.【Conclusion】The hyperspectral imaging technology is applicable for rapid evaluation of wilting degree of eggplant. The combination of SG pretreatment and PLS-DA discrimination model can achieve rapid，non-destructive， and accurate discrimination of wilting degree of eggplant seedlings after high temperature stress，thereby providing a feasible method for high-throughput identification of heat-tolerant eggplant phenotypes.