DETECTION OF NUTRIENT DEFICIENCY IN CROP PLANTS USING DEEP LEARNING MODEL

Jeba Faizah Rahman*, Md Anowar Parvez

Nutrient deficiency significantly affects plants growth and crop productivity. Hence, early detection is essential for effective agricultural management. This study aimed to develop a deep learning model for automated detection of healthy, potassium deficient (K) and nitrogen-potassium deficient (N-K) status in crop plants using leaf images. EfficientNetB0 architecture was used to build the model which was then trained and tested. The final model achieved 85.80% training accuracy and 76.61% test accuracy. Precision, recall and F1 score for test set was 76.03%, 76.49% and 75.93%%, respectively, indicating highly reliable prediction performance. Confusion matrix analysis for test set showed correct detection of 91 healthy, 72 potassium deficient and 122 nitrogen-potassium deficient crop leaves. Overall, the findings of the study indicate that the proposed model is effective for identifying nutrient deficiency in crop plants from leaf images.