Exploring Complementarity and Explainability in CNNs for Periocular Verification Across Acquisition Distances
This work addresses biometric verification for security applications, but it is incremental as it builds on existing CNN architectures and fusion techniques.
The study tackled periocular verification across distances by training three CNNs and using fusion methods, achieving a new state-of-the-art on the UBIPr database with substantial gains from network complementarity.
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.