A Vision-Language Foundation Model for Leaf Disease Identification

TL;DR

SCOLD is a novel vision-language model specifically designed for leaf disease identification, leveraging a large domain-specific dataset and soft-target contrastive learning to improve generalization and robustness in agricultural diagnostics.

Contribution

This work introduces SCOLD, a domain-specific, context-aware vision-language foundation model trained on a large plant leaf dataset, with soft-target contrastive learning to enhance fine-grained classification.

Findings

SCOLD outperforms existing models on zero-shot and few-shot tasks.

SCOLD demonstrates superior robustness and generalization in plant disease identification.

Ablation studies confirm the effectiveness of soft-target contrastive learning.

Abstract

Leaf disease identification plays a pivotal role in smart agriculture. However, many existing studies still struggle to integrate image and textual modalities to compensate for each other's limitations. Furthermore, many of these approaches rely on pretraining with constrained datasets such as ImageNet, which lack domain-specific information. We propose SCOLD (Soft-target COntrastive learning for Leaf Disease identification), a context-aware vision-language foundation model tailored to address these challenges for agricultural tasks. SCOLD is developed using a diverse corpus of plant leaf images and corresponding symptom descriptions, comprising over 186,000 image-caption pairs aligned with 97 unique concepts. Through task-agnostic pretraining, SCOLD leverages contextual soft targets to mitigate overconfidence in contrastive learning by smoothing labels, thereby improving model generalization and robustness on fine-grained classification tasks. Experimental results demonstrate that SCOLD outperforms existing vision-language models such as OpenAI-CLIP-L, BioCLIP, and SigLIP2 across several benchmarks, including zero-shot and few-shot classification, image-text retrieval, and image classification, while maintaining a competitive parameter footprint. Ablation studies further highlight SCOLD's effectiveness in contrast to its counterparts. The proposed approach significantly advances the agricultural vision-language foundation model, offering strong performance with minimal or no supervised fine-tuning. This work lays a solid groundwork for future research on models trained with long-form and simplified contexts, tasks involving class ambiguity, and multi-modal systems for intelligent plant disease diagnostics. The code for this study is available at https://huggingface.co/enalis/scold