Insect infestation in rice stock is a significant issue in rice exporting business, resulting in the loss of product quality, nutrient as well as the economic losses. However, detecting the insect contamination with the traditional sorting techniques were destructive, inaccurate, time consuming and unable to detect the internal insect infestation. This study used near infrared (NIR) spectroscopy for obtaining the absorbent spectra from the insect contamination in two kinds of rice samples, Milled Hommali rice (MHR) and Brown Hommali rice (BHR). The mathematical methods of partial least squares (PLSs) regression and singular value de-composition (SVD) were employed to construct the predicting model. The statistical analysis results, R2 , RMSEP, RPD and bias, concluded that the predictive models from PLS for MHR and BHR were 0.95 and 0.90, 0.014 and 0.019, 4.79 and 3.11, as well as 0.007 and 0.008, respectively; while the statistical analysis results from SVD for MHR and BHR were 0.97 and 0.96, 0.012 and 0.013, 5.71 and 5.39, as well as 0.003 and 0.002, respectively. It showed that SVD technique performed better than PLS technique which shows that using the advantage of SVD technique required less amounts of wave numbers for predicting and was possible to construct the low cost handheld equipment for detecting the insects in rice samples.