The prediction of the maturity date of leafy greens in a planting environment is an essential research direction of precision agriculture. Real-time detection of crop growth status and prediction of its maturity for harvesting is of great significance for improving the management of greenhouse crops and improving the quality and efficiency of the greenhouse planting industry. The development of image processing technology provides great help for real-time monitoring of crop growth. However, image processing technology can only obtain the representation information of leafy greens, and it is difficult to describe the causal mechanism of environmental factors affecting crop growth. Therefore, a framework combining an image processing model and a crop growth model based on causal inference was proposed to predict the maturity of leafy greens. In this paper, a deep convolutional neural network was used to classify the growth stages of leafy greens. Then, since some environmental factors have causal effects on the growth rate of leafy greens, the causal effects of various environmental factors on the growth of leafy greens are obtained according to the data recorded by environmental sensors in the greenhouse, and the prediction results of the maturity of leafy greens in the study area are obtained by combining image data. The experiments showed that the root mean square error (RMSE) was 2.49 days, which demonstrated that the method had substantial feasibility in predicting the maturity for harvesting and effectively solved the limitations of poor timeliness of prediction. This model has great application potential in predicting crop maturity in greenhouses.

Phenotype analysis of leafy green vegetables in planting environment is the key technology of precision agriculture. In this paper, deep convolutional neural network is employed to conduct instance segmentation of leafy greens by weakly supervised learning based on box-level annotations and Excess Green (ExG) color similarity. Then, weeds are filtered based on area threshold, K-means clustering and time context constraint. Thirdly, leafy greens tracking is achieved by bipartite graph matching based on mask IoU measure. Under the framework of phenotype tracking, some time-context-dependent phenotype analysis tasks such as growth monitoring can be performed. Experiments show that the proposed method can achieve 0.95 F1-score and 76.3 sMOTSA (soft multi-object tracking and segmentation accuracy) by using weakly supervised annotation data. Compared with the fully supervised approach, the proposed method can effectively reduce the requirements for agricultural data annotation, which has more potential in practical applications.