Table 6 Comparison with previous studies.
Author and year | Method used | Dataset source | Accuracy (%) | Identified gaps | Contribution |
|---|---|---|---|---|---|
Pandey et al. (2023)13 | Deep Convolutional Neural Network (CNN) | 2293 images of diseased leaves | 97.98 | Model verification lacking | High accuracy in disease detection with deep CNN |
Rai & Pahuja (2023)14 | Improved Deep CNN | Cotton leaf images | 97.98 | Model deployment issues | Achieved high accuracy in leaf disease classification |
Senthil Pandi et al. (2024)18 | Inception-V3 Transfer Learning (ITL-CHB) | Black Gram disease dataset | 98 | Data imbalance in training | Improved disease classification accuracy using transfer learning |
Stephen et al. (2024)19 | Deep Learning (CNN) | Big data-based cotton monitoring system | 93.9 | Rare disease detection issues | Big data-driven system for improved plant health monitoring |
Arathi & Dulhare (2023)20 | DenseNet-121 with Transfer Learning | Cotton leaf disease dataset | 91 | Requires fine-tuning for rare diseases | Enhanced cotton disease classification with transfer learning |
Vibhute & Kale (2023)21 | Machine Learning (PLSR, SVM, SAM) | Soil data (spectroradiometer, satellite images) | 95 | Limited generalizability to other crops | Machine learning models for soil management in cotton farming |
Paul et al. (2024)25 | YOLOv8 for Object Detection | Capsicum harvesting dataset | 96.7 | Requires extensive training data | Automated harvesting with YOLOv8 for detection and real-time application |
Proposed method (ours) | MobileNet, VGG19,CNN, LSTM, RNN, TLA+ | Hybrid (public + field data) | 98.7 | Addressed requirments correctness & model deployment | Mathematical verification of requirements, multiple class cotton disease prediction |
