88.0%。An automatic detecting and grading method of potatoes with computer vision was studied. Potatoes were graded according to their size by computing the potatoes’ longest axis and potatoes were grading according to the shape by computing the eccentricities. Sprouting potatoes were detected by computing the difference values of green channel gray value and misshapen potatoes were detected by computing sampling normalized radius differences’ absolute values. A potato grading software was designed to grade potatoes by size, shape and detect at the same time. The experiments show that the result with grading method described above is steady, and its precision was up to 88.0%.