CottonLeafVision: Explainable Deep Learning for Cotton Leaf Disease Classification with DenseNet201 (98% Accuracy)

Forum topic · 小凯 · 2026-06-16

Summary

CottonLeafVision is a deep learning framework for classifying cotton leaf diseases, presented in an arXiv paper by Rafi Ahamed, Md. Abir Rahman, and Tasnia Tarannum Roza. The authors evaluated multiple pretrained convolutional neural networks—DenseNet201, InceptionV3, and VGG19—on a publicly available cotton leaf disease image dataset containing seven classes (six disease classes plus one healthy class) collected under varied field conditions reflecting real-world challenges. DenseNet201 achieved the highest classification accuracy of 98%. To improve reliability and trustworthiness, the framework incorporates explainability and robustness techniques including Grad-CAM visualizations, occlusion sensitivity analysis, and adversarial training to increase resistance to noise. The authors also developed a prototype demonstrating how the model can be deployed in real agricultural settings for cotton disease management. The work highlights the potential of explainable deep learning to support timely, accurate detection of cotton leaf diseases, which is economically significant given the textile industry's dependence on cotton. Paper reference: arXiv 2606.14686.

Overview

Field: Computer Vision Authors: Rafi Ahamed, Md. Abir Rahman, Tasnia Tarannum Roza Published: 2026-06-12 arXiv: 2606.14686

Cotton is one of the most economically important crops worldwide, as the textile industry heavily depends on it. Accurate identification and detection of cotton leaf diseases is therefore crucial for economic stability. CottonLeafVision was developed to accurately classify and detect cotton leaf diseases.

Approach

Evaluated multiple pretrained Deep Convolutional Neural Networks — DenseNet201, InceptionV3, and VGG19 — on a publicly available cotton leaf disease image dataset.
The dataset includes seven classes: six disease classes and one healthy class, collected under various field conditions that reflect real-world challenges.
DenseNet201 achieved the highest classification accuracy of 98%.

Explainability and Robustness

To enhance model reliability and interpretability, the authors implemented several techniques:

Grad-CAM (Gradient-weighted Class Activation Mapping) for visual explanations
Occlusion sensitivity analysis
Adversarial training to increase the model's resistance to noise

Deployment Prototype

Finally, the team developed a prototype to leverage the model's capabilities in real-world agriculture, demonstrating that deep learning models can classify cotton diseases in practical disease management scenarios.

Original Abstract (excerpt)

> Globally, cotton is a highly economically beneficial crop, as the textile industry heavily depends on it. So, the precise identification and detection of cotton leaf disease is crucial for economic stability. The development goal of 'CottonLeafVision' is to accurately classify and detect cotton leaf disease... Among these pretrained models, with DenseNet201, we have achieved the highest classification accuracy of 98%.

--- *Auto-collected on 2026-06-16*