Germplasm is the foundation of agriculture and a national strategic resource. The quality of seeds affects the production scale and potential of the overall planting industry. There are some problems in Chinast seed industry, such as incomplete and opaque industrial chain information and imperfect and unsafe seed traceability systems. This paper integrates blockchain technology into traditional seed information traceability and proposes a seed quality traceability model for distributed data storage. This paper uses Ethereum, Internet of Things, and terminal applications to design a threelayer architecture: application layer, Internet of things layer, and data storage layer. According to the characteristics of Ethereum and the requirements of system data storage and query, a distributed multidatabase storage mode is designed. The system stores the data information of each link from seed production to the circulation department. Combined with the characteristics of blockchain data information that cannot be tampered with, high transparency and traceability, it ensures that all data in the seed supply chain will not be changed or hidden by others to realize accurate traceability and rapid identification of seed quality. The system uses the smart contract algorithm to directly hit the source of the seed data to ensure the quality and safety of the purchased seeds and to provide information technology and service guarantees for safeguarding the rights and interests of seed consumers.

It is an important way to solve the shortage of rural labor force and reduce the cost of fruit picking to realize mechanized intelligent picking in orchards. Accurate identification of fruit in orchards is the key technology. We took jujube as the research object. In order to establish a maturity identification model suitable for multiple varieties and strong practicability, the jujube fruits of many varieties in natural environment were divided into mature fruit, immature fruit and ripe fruit, semired fruit and immature fruit labeling methods, and four recognition models based on YOLO V3, YOLO V4, YOLO V4-Tiny and Mobilenet-YOLO V4-Lite were established. The study showed that both YOLO V3 and YOLO V4-Tiny models in the YOLO algorithm could be applied to the two labeling methods, and the verification set mAP was about 94%, which proved that the YOLO algorithm could effectively identify the maturity of jujube fruits.

In order to realize the fast and accurate optimization of corn seeds, a watershed algorithm combined with convolutional neural network was proposed to detect the quality of corn seeds with different quality as the research object. Firstly, the watershed algorithm was used to divide the single corn seed, and then the quality of each seed was classified by the convolutional neural network model. According to the position of the single seed obtained by the watershed algorithm, the results were labeled in the image to realize the quality detection of seeds. The improved InceptionV3 model was used to test the seeds. The test results showed that the average accuracy rate, the average recall rate and the F1 value (harmonic average evaluation) of the two kinds of seeds with good quality and defects were 94.18%, 94.61% and 94.39%. Meanwhile, in order to highlight the performance of the convolutional neural network model, the results were compared with the traditional machine learning method, and the F1 value of the convolutional neural network model was 20.39% higher than that of the LBP+SVM model.

This paper aims to solve the problems of low recognition rate of potato leaf diseases and difficulty in localization of late blight spots in the natural environment based on potato leaf images collected in the field environment. Firstly, the potato leaf disease was identified using five neural network models: ALEXNET, VGG16, InceptionV3, RESNET50, and MobileNet. The results show that the InceptionV3 model has the highest recognition accuracy, which can reach 98.00%. Secondly, the late blight spots of potato leaves were detected, and an improved CenterNet-SPP model was proposed. The model obtains the center point of the object through the feature extraction network and then obtains the image information such as the offset of the center point and the size of the target through the regression of the center point. The mAP of the trained model under the verification set was up to 90.03%.Using F ₁ as the evaluation value analysis, compared with other objective detection models, the CenterNet-SPP model has the best effect, whose accuracy is 94.93%, recall rate is 90.34%, F ₁ value is 92.58%, and the average detection time of an image is 0.10 s.This paper provides a relatively comprehensive deep learning algorithm and model research basis for potato disease recognition and detection in the natural environment.