A Deep Learning Approach for Diagnosing Nutrient Deficiencies in Okra Leaves

Precision farming and production optimization depend on accurate nutrient deficit (ND) diagnosis.  In order to diagnose various problems in okra leaves, this study proposes a repeatable and comprehensible deep-learning pipeline that combines segmentation, adaptive picture improvement, handcrafted deep feature fusion, and a lightweight convolutional neural network (CNN) classifier. By simultaneously adjusting Gaussian and Poisson noise weights, a Dual Adaptive Weighted Kalman Filter (DAWKF) reduces noise in input images.  Segmentation is carried out via an Adaptive Weighted Optimizer with Otsu Thresholding (AWO-AOT) using bio-inspired adaptive migration methods that improve local thresholds.  Local Binary Patterns (LBP), FAST, and SIFT descriptors are used in Adaptive Fused Features (AFF), which feed a SIFT-guided CNN to learn spatial-texture representations. 2,030 agronomist-verified okra-leaf photos from various field sources make up the extended dataset. They are divided into 70/15/15 sections for training, validation, and testing, and they are assessed using five-fold cross-validation.  The suggested framework outperformed BM3D and PSO-Otsu baselines with 98.6 ± 0.3% accuracy and statistically significant gains (p < 0.05) in PSNR (+1.8 dB) and SSIM (+0.014). Grad-CAM explainability studies revealed important symptom regions associated with necrosis and leaf chlorosis.  The technique provides a repeatable mechanism for real-field agronomic decision assistance and is generalizable across lighting, camera, and field conditions.