Deep Learning-Based Fertilizer Recommendation System Using Nutrient Deficiency Prediction

Nutritional deficiencies in plant lead to stunted growth in plants, which can result in low crop productivity. Currently machine learning models are employed for detecting abnormalities in the plant leaves images but the advent of deep learning has greater potential for processing the images in terms of increased accuracy. In this paper convolution, neural network model was developed for diagnosis of deficiency in plant leaves using the images captured from digital camera. The training was performed with 20777 images with 11 classes of nutritional deficiencies. This study uses ResNet-50 Deep convolution neural network for processing plant leaves images and accurately recommend the appropriate nutrients to be provided to overcome the deficiency in the form of fertilizer recommendations. Compared to other models the accuracy of the proposed method is 98.18%, which outnumbers other methods.