Deep learning based detection of plant nutrient deficiency symptom and design of multi-layer greenhouse system

In view of the current situation that the per capita cultivated land in agriculture is insufficient, the yield per mu is reduced and the degree of intelligence in the traditional greenhouse is low, A multi-layer intelligent farm with deep learning algorithm is proposed. Firstly, the overall design of the system is described in detail, including multi-layer greenhouse frame, water, fertilizer and medicine integrated machine, two-dimensional interpolation inspection robot, ventilation fan, fluorescent lamp, circulating water curtain; Secondly, the electrical system and cloud control system are designed. Finally, a deep learning network based on YoloV4-Tiny is installed on Raspberry PI to solve the image recognition problem of rose deficiency and insect pests. After a large number of experimental tests, it is found that the speed and accuracy of using YOLOV4-Tiny are improved compared with using YOLOV4. It solves the common problems of difficult to capture objects and slow recognition of features in other types of greenhouse systems, meets the requirements of ensuring good plant growth in different environments, and ensures the high quality and efficient operation of the greenhouse system.