Discrimination of plant root zone water status in greenhouse production based on phenotyping and machine learning techniques

Abstract

Plant-based sensing on water stress can provide sensitive and direct reference for precision irrigation system in greenhouse. However, plant information acquisition, interpretation, and systematical application remain insufficient. This study developed a discrimination method for plant root zone water status in greenhouse by integrating phenotyping and machine learning techniques. Pakchoi plants were used and treated by three root zone moisture levels, 40%, 60%, and 80% relative water content. Three classification models, Random Forest (RF), Neural Network (NN), and Support Vector Machine (SVM) were developed and validated in different scenarios with overall accuracy over 90% for all. SVM model had the highest value, but it required the longest training time. All models had accuracy over 85% in all scenarios, and more stable performance was observed in RF model. Simplified SVM model developed by the top five most contributing traits had the largest accuracy reduction as 29.5%, while simplified RF and NN model still maintained approximately 80%. For real case application, factors such as operation cost, precision requirement, and system reaction time should be synthetically considered in model selection. Our work shows it is promising to discriminate plant root zone water status by implementing phenotyping and machine learning techniques for precision irrigation management.

Crop Phenomics: Current Status and Perspectives

“…Corn leaf segment; graph to object converter; HSI TO GRAY converter, etc. Golzarian et al, 2011;Chen D. et al, 2014;Neilson et al, 2015;Amanda et al, 2016;Arend et al, 2016b;Cai et al, 2016;Guo et al, 2017;Liang et al, 2017;Majewsky et al, 2017;Meng et al, 2017;Neumann et al, 2017;Pandey et al, 2017;Parlati et al, 2017;Tomé et al, 2017; Crops within a 10-20 m × 110-200 m area can be monitored, which realizes the continuous, automatic, and high-throughput detection of crop phenotyping detection in field (Virlet et al, 2016;Sadeghi-Tehran et al, 2017). Meanwhile, the cablesuspended field phenotyping platform covering an area of ∼1 ha was also developed for rapid and non-destructive monitoring of crop traits (Kirchgessner et al, 2016).…”

Non-Invasive Sensing of Nitrogen in Plant Using Digital Images and Machine Learning for Brassica Campestris ssp. Chinensis L.

“…Based on the screened phenotypic features, three models, namely the random forests (RF), support vector regression (SVR), and neural network (NN), were developed to quantitatively predict the NNI of the pakchoi. For the model development, we referred to the study by Guo et al [43] with some modification. The three modeling algorithms were executed using the randomForest, nnet, and e1071 packages in R language (release 3.4.1), respectively.…”

“…This is particularly useful in the case of plant growth development, where it is challenging to efficiently model the holistic effect of genetic, physiological, phenotypic, agronomic, meteorological, and human factor features on the plant. ML technology has been applied to the identification [40,41], classification [42,43], quantification [44,45], and prediction [46,47] of stress phenotyping in plants. In summary, ML approaches are typically useful in situations where large amounts of data are available, relating inputs (e.g., phenotypic data) to output quantities of interest (e.g., NNI).…”

Optical Flow-Based Analysis of the Relationships between Leaf Wilting and Stem Diameter Variations in Tomato Plants

“…Other water stress estimation methods have been proposed utilizing plant images and environmental data measured without contacting plants [18][19][20][21][22][23][24][25][26]. Crucially, the contactless measurement allows new farmers to use the water stress estimation method without previous experience of taking such measurements.…”