Detection of Rice Seed Hole Seeding Amount Based on Deep Learning

Cultivating rice seedlings by using seedling tray sowing is an important part of large-scale industrialized rice planting. Rice seedling raising is to sow after germination. During the seed cultivation process, it will produce different characteristics and thus affect the seedling development or sowing effect. In order to improve the accuracy and efficiency of rice seed target detection of rice pot floppy disk hole seeding amount, this paper proposes a method of rice pot floppy disk hole seeding amount detection based on depth learning. The production data set should use the pictures taken from the seeded floppy disk on the top to train the network on the GPU computing platform by using the data set through the YOLOv5 neural network model. The analysis of the training results and the verification results show that the trained neural network model is a very suitable network model for the detection of the amount of rice seeds planted in holes in terms of real-time and other comprehensiveness. This method can automatically learn and extract the characteristics of rice seeds in the picture of the bowl floppy disk, and realize the real-time automatic detection of 0-7 and more rice seeds planted in holes in the bowl floppy disk, which is conducive to migration to embedded platforms.