Genetic algorithm enhanced deep learning with data augmentation for nitrogen and potassium deficiency detection in eggplant

Abstract

The quality of horticultural crops is significantly crucial for agricultural yield because of market demand, quality, and the priority of consumers. Macronutrients like nitrogen (N) and potassium (K) are crucial for the normal growth and development of crops. Thus, detecting nutritional deficiency in eggplant is very important for ensuring optimal growth and yield. The traditional approaches are time-consuming and require expert knowledge. The previously reported research in eggplant with a deep learning (DL) approach targeted disease detection and classification work. No work has been reported on eggplant nutritional deficiency detection using the genetic algorithm (GA) based tuning approach with data augmentation. This paper presents a YOLOv9 deep-learning model, optimized with a GA to find the best hyperparameters and data augmentation techniques to increase its robustness. The study used the OLID I dataset to detect nutritional deficiencies in eggplant leaves. The experimental results show that our approach achieved an accuracy of 94.52 %, mAP50 of 94.55 %, mAP50–95 of 93.23 %, Precision of 95.9 %, Recall of 92.8 %, and F1 Score of 94.32 %. These results suggest that the proposed approach is a significant step towards developing a practical application to support farmers in detecting nutrition deficiencies in the eggplant crop.