AI 中文总结
该研究构建了基于799张单粒榛子X射线图像的质量二分类基准，评估多种深度学习模型，发现专家重新标注模糊样本可提升性能，强调多拆分评估的必要性。
AI 中文摘要
无损X射线成像可揭示仅靠外观检查难以发现的榛子内部缺陷；然而，由于类别间细微的射线照相差异、显著的类别不平衡以及带标注数据有限，自动解读仍具挑战性。在此，我们提出一个基于799张分割后的单粒榛子X射线图像（尺寸为224×224像素，灰度图）的榛子质量二分类基准（健康vs缺陷），这些图像被归为101个采集单元。我们采用分组拆分-旋转协议，使用不同随机种子生成的5个数据拆分，对7种单模型配置和10种概率聚合集成模型进行评估。在验证集上选择决策阈值，并在验证集和测试集上确定性地评估性能。在专家重新标注的条件下，经二元交叉熵训练的卷积神经网络与冻结Swin Transformer的平均概率集成模型达到最高的平均平衡准确率（86.3%±1.8%，5个种子），另有多种集成模型提供了可比的性能。在不同方法中，观察到显著的拆分间变异性，表明在该数据集规模下，多拆分评估对于可靠的模型比较至关重要。专家对模糊样本的重新标注使所有17种评估方法的性能提升了2.8至8.1个百分点，而对拆分间方差的影响有限。结果凸显了深度学习在基于X射线的榛子质量自动评估方面的潜力，以及在小型、不平衡的农业成像数据集中严格评估和标注整理的重要性。
英文摘要
Non-destructive X-ray imaging can reveal internal hazelnut defects that are difficult to detect by external inspection alone; however, automated interpretation remains challenging because of subtle radiographic differences among classes, marked class imbalance, and limited annotated data. Here, we present a benchmark for binary hazelnut quality classification (healthy versus defective) based on 799 segmented single-kernel X-ray images (224 x 224 pixels, grayscale), grouped into 101 acquisition units. Seven single-model configurations and ten probability-aggregation ensembles were evaluated using a group-wise split-rotation protocol across five data splits generated using different random seeds. Decision thresholds were selected on the validation set, and performance was assessed deterministically on validation and test sets. Under the expert-reassessed annotation condition, the average-probability ensemble of the binary cross-entropy-trained convolutional neural network and frozen Swin Transformer achieved the highest mean balanced accuracy (86.3% +/- 1.8%, five seeds), with several other ensembles providing comparable performance. Across methods, substantial split-to-split variability was observed, indicating that multi-split evaluation is essential for reliable model comparison at this dataset scale. Expert reassessment of ambiguous samples improved the performance of all 17 evaluated methods by 2.8-8.1 percentage points, while having only a limited effect on cross-split variance. The results highlight both the potential of deep learning for automated X-ray-based hazelnut quality assessment and the importance of rigorous evaluation and label curation in small, imbalanced agricultural imaging datasets.
Comments26 pages (including 5 pages of supplementary material), 4 figures, 5 tables. Dataset available at https://doi.org/10.5281/zenodo.21739932
