The invention discloses a leaf vegetable yield prediction method which is aimed at overcoming a problem of limitation in a conventional yield in-vitro prediction and estimation method. The leaf vegetable yield prediction method comprises the following steps: in a first step, basic parameters are obtained; in a second step, data is stored and processed; in a third step, yield per plant is measured: (1) leaf mass prediction is carried out, and (2) yield per plant of lettuce is predicted via an equation m plant=11.297m leaf+48.827; in a fourth step, yield per area is measured: (1) ten plants of lettuce are chosen for measurement and yield prediction in an area being tested via use of a yield prediction model for a single plant of lettuce, and average mass per plant of lettuce in the area is obtained and used as an average yield value in the area being tested; (2) the average mass per plant of lettuce in the area is multiplied by the calculated number of lettuce plants in a lettuce image collected in the area, and an obtained product is lettuce yield data in the area; in an equation M=m average*N, M represents total yield of lettuce in the area, m average represents the average mass per plant of lettuce, and N represents the number of the lettuce plants in the area being tested.