Mobile Detection of Macronutrient Deficiencies in Lettuce Plants Using Convolutional Neural Network

As lettuce is one of the most widely consumed vegetable crops, detecting its deficiencies in macronutrients, nitrogen (N) and potassium (K), becomes an important task in ensuring its proper development and growth. This project has implemented a mobile application that detects nitrogen and potassium deficiencies in lettuce by utilizing lightweight Convolutional Neural Networks (CNN) models, specifically EfficientNetB0, MobilenetV3Small, and MobileNetV3Large, for the classification of macronutrient deficiencies. Additionally, the application includes a module for image processing that separates the crop from the background to improve classification accuracy. Furthermore, CNN models target low resource usage and can be used on mobile CPUs without compromising the accuracy in classifying macronutrient deficiencies. The mobile application was 81 % accurate in detecting whether the lettuce plants were healthy, nitrogen-deficient, or potassium-deficient.