EASE-PVNet: Robust Periocular Identity Verification Across Pre- and Post-Operative Facial Images

Identity verification across pre-operative and post-operative facial images remains a challenging task, particularly following eyelid surgery, where localized periocular changes can disrupt conventional face recognition systems. This research introduces a novel verification framework using an ensemble-based autoencoder-initialized siamese eye-region periocular verification network designed to remain resilient to surgically induced appearance variation. The proposed approach integrates anatomy-guided periocular normalization with a Siamese deep metric learning architecture initialized through unsupervised autoencoder pretraining, allowing the model to acquire periocular-specific representations prior to supervised learning. Robustness in this data-limited clinical setting is further enhanced through staged hard-negative mining, validation-weighted multi-seed ensemble learning, and bootstrap-based threshold calibration. Ensemble Grad-CAM is employed to provide visual explanations that support clinical interpretability. Experimental evaluation demonstrates strong and consistent performance, achieving recognition rates of 94.71% on training data, 96.77% on validation, and 96.08% on the test set, with an overall recognition rate of 95.24%. Compared to previously reported periocular verification methods which reported an overall recognition rate of only 91.8% under similar conditions. These results highlight the effectiveness and stability of the proposed framework for post-surgical periocular identity verification in clinical and forensic applications.