The potential of RGB camera for machine learning in non-destructive detection of nutrient deficiencies in apples

From a plant nutrition perspective, the appearance of colour changes and malformations on leaves and fruits usually indicates a nutrient imbalance in a complex and dynamic soil-plant-air system. Each nutrient deficiency symptom occurs differently on the plant. Observing such colour changes in the appearance of transformation could help fruit growers respond and prevent further nutritional problems. The aim of this research was to create a model that could be used as a tool for non-destructive detection of nutrient deficiencies on leaves. An RGB camera was used to manually record the occurrence of nutrient deficiencies in commercial apple orchards. Two hundred images were taken at each of five intervals during the day for several months of vegetation. The images were then processed in an annotation program (LabelImg) in which each leaf was classified into one of the following categories: healthy leaf or nitrogen, phosphorus, potassium, calcium, magnesium, iron, zinc, or manganese deficient. The data obtained from the latter program are used as training data which is used to build a model in the machine learning process. Machine learning is applied to a rover designed as a machine that records nutrient deficiencies with RGB cameras and drives autonomously through apple orchards. The training data were used as comparison points that enabled the machine to detect and classify nutrient deficiencies.