Objective: In order to solve the problem that temperature change during drying can detect the moisture content of Camellia oleifera seeds by visible/near infrared spectroscopy, a temperature modified Camellia oleifera seed moisture content detection model was proposed. Methods: Drying experiments were carried out at different temperatures (50,60,70 ℃) to collect spectral data. By acquiring the spectral data collected at different temperatures, the reasons why the temperature affected the spectrum were analyzed. Then, by comparing the three spectral preprocessing methods, using the Competitive Adaptive Reweighting(CARS)and Partial Least Squares Regression (PLSR) were used to establish the benchmark PLSR model at 60 ℃. Finally, the slope/bias method was used to correct the predicted values of external samples at 50 ℃ and 70 ℃, which greatly improved the precision and accuracy. Results: The coefficients of determination before and after correction at the two temperatures were 0.729 and 0.848, 0.763 and 0.862, respectively. The relative analytical error RPD values were 1.921 and 2.565, 2.054 and 2.692, respectively. Conclusion: The modified model could effectively improve the prediction accuracy, achieve good prediction effect, overcome the influence of temperature, and provide a new method to eliminate the influence of temperature when detecting the oil Camellia oleifera seed moisture content by visible/near infrared spectroscopy in the drying field.