探索卷积神经网络在虹膜周边验证中跨采集距离的互补性与可解释性 (Exploring Complementarity and Explainability in CNNs for Periocular Verification Across Acquisition Distances)

We study the complementarity of different CNNs for periocular verification at different distances on the UBIPr database. We train three architectures of increasing complexity (SqueezeNet, MobileNetv2, and ResNet50) on a large set of eye crops from VGGFace2. We analyse performance with cosine and chi2 metrics, compare different network initialisations, and apply score-level fusion via logistic regression. In addition, we use LIME heatmaps and Jensen-Shannon divergence to compare attention patterns of the CNNs. While ResNet50 consistently performs best individually, the fusion provides substantial gains, especially when combining all three networks. Heatmaps show that networks usually focus on distinct regions of a given image, which explains their complementarity. Our method significantly outperforms previous works on UBIPr, achieving a new state-of-the-art.

翻译：本研究基于UBIPr数据库，探究了不同卷积神经网络（CNN）在不同距离下进行虹膜周边验证的互补性。我们使用VGGFace2数据集中的大量眼部裁剪图像，训练了三种复杂度递增的网络架构（SqueezeNet、MobileNetv2和ResNet50）。通过余弦相似度和卡方（chi2）度量分析性能，比较了不同网络初始化方式，并采用逻辑回归进行分数级融合。此外，我们利用LIME热力图和Jensen-Shannon散度比较了各CNN的注意力模式。结果显示，虽然ResNet50单独使用时始终表现最佳，但融合策略（尤其是三网络组合）带来了显著性能提升。热力图表明不同网络通常聚焦于图像的不同区域，这解释了它们的互补性。本方法在UBIPr数据库上显著超越先前工作，达到了新的最优性能水平。