A100: A Hybrid Vision Foundation Model for Agricultural Plant Image Analysis

📄 arXiv: 2610.22991v1 📥 PDF

作者: Mayur Pratim Das, Ajai Dubey, Aksshay Mathew P, Sana Afreen, Rishi Kumar Soni, Deepak Meena, Pratibha Kumari, Mukesh Saini

分类: cs.CV

发布日期: 2026-10-16

备注: 11 pages, 5 figures, 5 tables

💡 一句话要点

提出A100以解决农业植物图像分析中的数据不足问题

🎯 匹配领域: 支柱二：RL算法与架构 (RL & Architecture) 支柱九：具身大模型 (Embodied Foundation Models)

关键词: 农业视觉模型 自监督学习 混合CNN-Transformer 作物监测 疾病诊断 植物表型分析 标签效率 深度学习

📋 核心要点

  1. 现有农业视觉模型在作物监测和疾病诊断中面临数据不足的问题，尤其是在发展中国家。
  2. A100模型通过自监督学习在大量未标记图像上进行预训练，结合CNN和Transformer架构以提升特征提取能力。
  3. 实验结果显示，A100在多个基准数据集上超越现有模型，尤其在有限标注数据下表现出色，具有良好的标签效率。

📝 摘要（中文）

在农业中，准确的视觉模型用于作物监测、疾病诊断和植物表型分析，但由于缺乏专家注释的数据，尤其是在发展中国家，这一应用受到限制。本文提出了A100，一个农业视觉基础模型，利用自监督学习在300万张未标记的图像上进行预训练，涵盖100多种作物。A100采用混合CNN-Transformer骨干网络，结合MBConv阶段进行局部特征提取和Transformer阶段进行全局上下文表示。通过DINO自蒸馏、iBOT掩码补丁重建和KoLeo均匀性正则化的集成，A100在农业应用中表现出比ImageNet-12K监督表示更优的特征。经过全量微调，A100在32类作物识别和44类疾病检测上的表现分别达到99.90%和99.02%。

🔬 方法详解

问题定义：本文旨在解决农业植物图像分析中缺乏专家注释数据的问题，现有方法在数据稀缺情况下表现不佳，限制了作物监测和疾病诊断的准确性。

核心思路：A100通过自监督学习在300万张未标记图像上进行预训练，结合混合CNN-Transformer架构，旨在提升特征提取的局部与全局能力，从而提高农业应用的准确性。

技术框架：A100的整体架构包括MBConv阶段用于局部特征提取和Transformer阶段用于全局上下文表示，采用DINO自蒸馏、iBOT掩码补丁重建和KoLeo均匀性正则化等技术进行预训练。

关键创新：A100的主要创新在于其混合架构和自监督学习策略，使得模型在没有微调的情况下，便能在农业应用中超越传统的ImageNet-12K监督表示。

关键设计：A100采用冻结骨干网络的策略，确保在作物分类中达到95.37%的准确率，并在GroMo25表型基准上减少10.4%的叶片计数误差，展示了其在分类和回归任务中的有效性。

🖼️ 关键图片

fig_0
fig_1
fig_2

📊 实验亮点

A100在多个基准测试中表现出色，特别是在PlantDoc数据集上，其准确率达到98.01%，并且在仅使用1%的标注数据时，超越了现有的农业基础模型FoMo4Wheat和LeafVision超过14个百分点，显示出显著的性能提升和标签效率。

🎯 应用场景

A100模型在农业领域具有广泛的应用潜力，能够有效支持作物监测、疾病诊断和植物表型分析。其高效的标签利用率使得在数据稀缺的情况下，农业专业人士能够快速适应和应用该模型，推动农业智能化的发展。未来，A100有望在全球范围内促进精准农业的实施。

📄 摘要（原文）

The use of accurate vision models for crop monitoring, disease diagnosis, and plant phenotyping in agriculture is hindered by the lack of expertly annotated data, especially in developing countries. We propose A100, an agricultural vision foundation model that is pre-trained using self-supervised learning on 3 million unlabeled images from fields covering more than 100 crop species. A100 uses a hybrid CNN-Transformer backbone that fuses MBConv stages for local feature extraction with Transformer stages for global contextual representation. A100 uses pre-training with the integration of DINO self-distillation, iBOT masked patch reconstruction, and KoLeo uniformity regularization, resulting in features that are better than ImageNet-12K supervised representations for agricultural applications without any fine-tuning. By utilizing a frozen backbone, the accuracy for crop classification of A100 is 95.37%, which is +2.56% better than ImageNet-12K, and reduces the leaf count error by 10.4% on the GroMo25 phenotyping benchmark, proving that the same backbone can be used for classification and regression tasks. After full fine-tuning, the performance of A100 is 99.90% on 32-class crop identification and 99.02% on 44-class disease detection. On the difficult PlantDoc dataset, A100 outperforms all baselines, such as CvT-13 (64.89% against 63.56%). Remarkably, A100 obtains 98.01% accuracy on plant classification using just 1% of the labelled data (about 284 images per category), showing great label efficiency for use cases with limited labeling. Under an identical protocol, A100 also outperforms the publicly released agricultural foundation models FoMo4Wheat and LeafVision by more than 14 points at 1% labels. We plan to publicly release our pretrained model to allow agricultural professionals to adapt A100 for various plant analysis applications using limited annotations.