A deep learning mobile application for tomato leaf nutrition deficiency identification with YOLOv8 and enhanced...

Nutritional deficiencies in crops lead to significant yield losses. Early and accurate detection of these deficiencies is crucial for effective intervention, as it enables timely corrective measures, minimizes crop damage, and ensures optimal productivity. Traditional methods for identifying deficiencies generally rely on manual inspection of the leaves, which is time-consuming and prone to errors. To address this challenge, we propose a deep learning (DL)-based approach that uses the latest advancements in object detection model YOLOv8 to detect and locate the affected tomato leaves. This paper presents a DL approach for detecting nutritional deficiencies in tomato leaves using the YOLOv8 model and a layered augmentation scheme. The augmented dataset allows the model to identify subtle signs of deficiencies better, leading to more accurate predictions. Our method was evaluated based on key performance metrics such as accuracy, memory usage, and mAP50. The model achieved an mAP@0.50 of 92.7 % and an mAP@0.50-0.95 of 89.1 %, with outstanding precision of 89.1 %, recall rate of 83.1 %, and a balanced F1 score of 89.5. The results demonstrate that the proposed framework achieves superior performance in detecting nutritional deficiencies when compared to earlier models in terms of mAP, accuracy precision, recall and F1 score. The combination of advanced object detection capabilities and an augmented dataset makes this approach valuable for precision agriculture, enabling timely and targeted interventions to improve crop health and productivity. Also, an Android app has been developed for real-time application of the approach.