Abstract: To reduce the labor cost of quality grading, we established a unified grading standard, improved the accuracy and efficiency of sorting and packing of Zizania, and developed an automatic sorting and packing equipment of Zizania based on machine vision. Quality grading of Zizania was achieved by deep learning technology to obtain phenotypic characteristics of Zizania such as size, shape, color, diseases, and pests. The equipment was composed of a control system, a feeding module, a quality grading module, and a sorting and packing module. The functions of automatic dividing, transmission, quality grading, sorting, and packaging were accomplished. Dividing and transferring mechanisms were designed to constrain Zizania to be separately transmitted orderly and steady. The 4-DOF collaborative control of manipulator ensured sorting and packaging of Zizania orderly. The testing result showed that the accuracy of quality grading of Zizania was 95.62%, and the efficiency of sorting and packing was up to 3 seconds per Zizania, with good stability. The results showed that the equipment could complete the sorting and packing of Zizania automatically and accurately. The equipment could also provide technical support for the transformation of “human labor replaced with machines” and for the promotion of agricultural intelligence level.