Aiming at the problems of cavity and single hole rate in rice seedling planting, in order to further improve the qualification rate of 2~3 grains per holes, a visual inspection technology of seeding quality was presented, and an efficient intelligent reseedling decision system was designed. The visual collection and image processing system were used to check the quality of seeding, and a population distribution database was established. The convolution calculation method was adopted to intelligently decide the reseeding scheme of the cavity and the single hole, and the dynamic coordinates of the replanted acupoints were positioned. The relevant parameters of seed suction were calculated to ensure its reliable performance. Finally, the combination of the servo motor and the pneumatic combined suction needle was used to form the reseeding mechanism, which realized an accurate, efficient and continuous dynamic reseeding process. According to the statistics of the test results, the average reseeding time per hole was 1.51s when the rice seed replenishment rate was 2%, and the intelligent seed replenishment system can meet the production requirements of 420 dishes per hour. The intelligent replanting decision greatly improved the working efficiency of the replanting system, and which can provide a theoretical basis for lowseeded reseeding technology of rice, vegetables, flowers and other crops in the factory. 