cs.CVMar 18, 2026

VISER: Visually-Informed System for Enhanced Robustness in Iris Presentation Attack Detection

Authors: Byron Dowling, Jacob Piland, Eleanor Frederick, Christopher Sweet, Adam Czajka

Organizations: †Department of Computer Science and Engineering, University of Notre Dame, Notre Dame, Indiana 46556, USA · ‡Center for Research Computing, University of Notre Dame, Notre Dame, Indiana 46556, USA

Read PDF
Ask AI
Cite

Copy citation

DOINot available for this paper

Abstract

Human perceptual priors have shown promise in saliency-guided deep learning training, particularly in the domain of iris presentation attack detection (PAD). Common saliency approaches include hand annotations obtained via mouse clicks and eye gaze heatmaps derived from eye tracking data. However, the most effective form of human saliency for raising generalization to unknown attack classes in iris PAD remains under-explored. In this paper, we conduct a series of experiments comparing hand annotations, eye tracking heatmaps, segmentation masks, and foundation model embeddings to a state-of-the-art deep learning-based baseline on the task of unknown attack type classification for iris PAD. Results in a leave-one-attack-type out paradigm indicate that denoised eye tracking heatmaps show the best generalization improvement over cross entropy in Attack Presentation Classification Error Rate (APCER) at Bona Fide Presentation Classification Error Rate (BPCER) of 1%. Along with this paper, we offer trained models, code, and saliency maps for reproducibility and to facilitate follow-up research efforts.

Figures & tables

Figure 1: Experimental pipeline and paper contributions. We start with a DenseNet121 architecture with no saliency as our baseline. Different forms of saliency are then tested on the architecture using saliency-guided training. Next foundation model embeddings are tested in SVM-Linear, SVM-RBF, and CNN classifiers. Finally, AP/BPCER 1% results are expressed as +/- improvement over the DenseNet121 baseline.
Figure 2: Sample salience types captured for the same iris image.
Figure 3: Example of applying HDBSCAN to de-noise the eye tracking data. Clusters made up of valid fixations are distinguished by color and larger marks denote longer fixations, not fixation area. Black crosses indicate fixations marked as noise excluded from the final saliency map.
MethodPrintoutDiseasedPost MortemSyntheticContacts + PrintTextured ContactArtificialAverage Δ\Delta *
Baseline Performance
DenseNet (XENT)0.55400.93770.83210.98310.23370.91670.98680.7777
Segmentation Masks
Baseline Saliency-based-0.0522+0.0211+0.0615+0.0094+0.0920-0.0587+0.0088+0.0117
Hand Annotations
High Entropy+0.0256+0.0088+0.0167+0.0169+0.0623-0.0293+0.0000+0.0144
Table 1: Compact Δ\Delta APCER @ BPCER=1% vs XENT Baseline ( ↓\downarrow ) results are presented.
Table 2: Comparison of Saliency-Augmented Training and Foundation Model Performance.
More in this topic

Visual Saliency

Visual saliency studies which image and video regions attract attention or matter for a task, and how to predict or explain that importance.

See all 31 papers

Explore similar work

CardsList

  1. Diff-IrisPAD: Modeling Bona Fide Iris Distributions for Unsupervised Presentation Attack Detection

    Oct 21, 2026Geetanjali Sharma, Harsh Yadav, Tanmay Deshpande +3Biometric IdentificationReconstruction-Based Anomaly Detection

    Indian Institute of Technology Mandi, Himachal Pradesh, India · SAFE Center, Norwegian University of Science and Technology (NTNU), Gjøvik, Norway

  2. Compact Vision Models for Iris Presentation Attack Detection under Presentation Attack Instrument Shift and Environmental Degradation

    Sep 17, 2026Athanasios Angelakis, Marta Gomez-BarreroDistribution Shift Robustness

    BioML Lab, RI CODE, UniBw, Munich, Germany

  3. A Systematic Failure Analysis of Vision Foundation Models for Open Set Iris Presentation Attack Detection

    May 18, 2026Rahul Anand, Siddharth Singh, Dileep A D +2Vision Foundation ModelsOpen-Set Recognition

    Indian Institute of Technology, Dharwad, India. · SAFE Center, Norwegian University of Science and Technology (NTNU), Norway. · Indian Institute of Information Technology Dharwad, India.