Deep learning model integrating YOLOSeg v8 for accurate leaf segmentation and a classification model for nutrient deficiency identification in crops. YOLOSeg v8 effectively estimates bounding boxes, objectness scores, and segmentation masks, ensuring precise leaf isolation. The proposed framework achieving a Dice coefficient of 97.18%, precision of 95.86%, recall of 98.20%, and F1-score of 97.18%. The model also demonstrated strong object detection with a Box Accuracy of 97.80% and mAP50 of 98.00%. Despite the computational intensity, with a prediction time of 13.5 milliseconds per image, the model is well-suited for real-time agricultural applications. Additionally, the custom CNN model achieved 82.40% test data accuracy and 82.89% precision, enhancing the reliability and ac-curacy of nutrient deficiency detection for sustainable agriculture. These results validate the model’s robustness for intelligent agriculture systems. |
*** Title, author list and abstract as submitted during Camera-Ready version delivery. Small changes that may have occurred during processing by Springer may not appear in this window.