ConMamba: Contrastive Vision Mamba for Plant Disease Detection

2025-06-03

AI Summary

To address key bottlenecks in plant disease detection—namely, scarcity of labeled data, high computational overhead, and weak modeling of long-range dependencies—this paper proposes a lightweight and efficient self-supervised framework. Methodologically, it adopts an unsupervised pretraining followed by supervised fine-tuning paradigm. Its core contributions are: (1) a novel vision Mamba encoder integrating bidirectional state space models (SSMs), substantially enhancing long-range visual representation learning; and (2) a dynamically weighted two-level contrastive loss that adaptively aligns local and global features. Evaluated on three benchmark datasets, the framework consistently outperforms state-of-the-art methods, achieving average improvements of 2.3–4.1% in accuracy and F1-score, alongside 37–58% faster inference speed—demonstrating synergistic optimization of both precision and efficiency.

Technology Category

• Machine Learning: Unsupervised & Self-Supervised Learning
• Computer Vision: Learning & Optimization for CV
• Search and Optimization: Learning to Search

Application Category

• Search and Retrieval-Augmented AI: Web learning to rank, online learning, and counterfactual learning for ranking
• Web Mining and Content Analysis: Large pretrained models with web data
• Graph Algorithms and Modeling for the Web: Efficient manipulation of static and dynamic Web-related graphs

Abstract

Plant Disease Detection (PDD) is a key aspect of precision agriculture. However, existing deep learning methods often rely on extensively annotated datasets, which are time-consuming and costly to generate. Self-supervised Learning (SSL) offers a promising alternative by exploiting the abundance of unlabeled data. However, most existing SSL approaches suffer from high computational costs due to convolutional neural networks or transformer-based architectures. Additionally, they struggle to capture long-range dependencies in visual representation and rely on static loss functions that fail to align local and global features effectively. To address these challenges, we propose ConMamba, a novel SSL framework specially designed for PDD. ConMamba integrates the Vision Mamba Encoder (VME), which employs a bidirectional State Space Model (SSM) to capture long-range dependencies efficiently. Furthermore, we introduce a dual-level contrastive loss with dynamic weight adjustment to optimize local-global feature alignment. Experimental results on three benchmark datasets demonstrate that ConMamba significantly outperforms state-of-the-art methods across multiple evaluation metrics. This provides an efficient and robust solution for PDD.

Problem

• Reducing reliance on costly annotated datasets for plant disease detection
• Addressing high computational costs in self-supervised learning methods
• Improving long-range dependency capture and local-global feature alignment

Innovation

• Vision Mamba Encoder for long-range dependencies
• Dual-level contrastive loss for feature alignment
• Dynamic weight adjustment in loss function