标题： DDD：用于植物疾病诊断的判别难度距离 显示英文标题

标题： DDD: Discriminative Difficulty Distance for plant disease diagnosis

摘要： 最近使用机器学习（ML）进行植物疾病诊断的研究强调了由于数据划分不当而导致的诊断性能被高估的问题，其中训练集和测试集的数据来源于同一来源（领域）。 植物疾病诊断是一个具有挑战性的分类任务，其特点包括细粒度、模糊的症状以及每个领域内图像特征的广泛变异性。 在本研究中，我们提出了判别难度距离（DDD）的概念，这是一种新的度量标准，旨在量化训练集和测试集之间的领域差距，同时评估测试数据的分类难度。 DDD为识别训练数据中的不足多样性提供了有价值的工具，从而支持构建更加多样和稳健的数据集。 我们研究了在不同数据集上训练的多个图像编码器，并检查了通过编码器生成的低维表示测量的数据集之间的距离是否适合作为DDD度量标准。 该研究使用了来自27个领域的四种作物和34种疾病类别的244,063张植物疾病图像。 结果表明，即使测试图像来自与编码器训练时不同的作物或疾病，将其纳入可以构建一个与独立开发的疾病分类器所指示的诊断难度高度相关的数据集距离度量。 与仅在ImageNet21K上预训练的基础编码器相比，相关性提高了0.106至0.485，最高达到0.909。 显示英文摘要

摘要： Recent studies on plant disease diagnosis using machine learning (ML) have highlighted concerns about the overestimated diagnostic performance due to inappropriate data partitioning, where training and test datasets are derived from the same source (domain). Plant disease diagnosis presents a challenging classification task, characterized by its fine-grained nature, vague symptoms, and the extensive variability of image features within each domain. In this study, we propose the concept of Discriminative Difficulty Distance (DDD), a novel metric designed to quantify the domain gap between training and test datasets while assessing the classification difficulty of test data. DDD provides a valuable tool for identifying insufficient diversity in training data, thus supporting the development of more diverse and robust datasets. We investigated multiple image encoders trained on different datasets and examined whether the distances between datasets, measured using low-dimensional representations generated by the encoders, are suitable as a DDD metric. The study utilized 244,063 plant disease images spanning four crops and 34 disease classes collected from 27 domains. As a result, we demonstrated that even if the test images are from different crops or diseases than those used to train the encoder, incorporating them allows the construction of a distance measure for a dataset that strongly correlates with the difficulty of diagnosis indicated by the disease classifier developed independently. Compared to the base encoder, pre-trained only on ImageNet21K, the correlation higher by 0.106 to 0.485, reaching a maximum of 0.909.