Paddy rice is one of the most crucial crops worldwide. In Taiwan, 3500 paddy seed cases (4000 grains/case) are manually inspected every year. However, red paddy seeds are abnormal and must be inspected. Because red paddy is covered with paddy hull, it cannot be recognized according to their appearance. In this study, we developed an automatic inspection system with an inlet-outlet mechanism, machine vision, and a control system. The machine vision including the camera, infrared sensor, backlight source, and inspecting software is used to recognize the red paddy with hull. The system can inspect red paddy seeds through machine vision. Our results indicated that the proposed system can inspect the transmittance of paddy seeds and then classify them as either “good” or “red” (not good), with a classification accuracy of 91.0%. The developed system is a sensor for red paddy efficiently.