The invention discloses a corn real-time detection and grading machine vision system. The system consists of a power and transmission mechanism, a computer vision identification mechanism, a single granulating device and an automatic grading mechanism, wherein grains enter from a feed port, speeds of the grains are controlled by the power and transmission mechanism; the grains are conveyed one byone by virtue of the single granulating device; pictures of the single granulated grains are collected by the computer vision identification mechanism; the collected pictures are subjected to pictureprocessing to acquire characteristic information of each grain, such as shape, color, size, surface smoothness, surface defect and the like; and the grade of the grain is determined in mode identification. In the whole process, each grain is tracked, when the grade identified grain reaches a discharge area, the automatic grading mechanism of grains is controlled by a control module of the computer vision identification mechanism to realize that the identified grain automatically falls in a collection case from the discharge port. Therefore, the level of automatically detecting and grading thegrains is improved.