Logo
ScienceStack
Table of Contents
Fetching ...
Sign In

ChatGPT Meets Iris Biometrics

Parisa Farmanifard, Arun Ross

TL;DR

The paper investigates whether a multimodal LLM, GPT-4, can analyze iris images for biometric recognition and compares its performance to Google's Gemini Advanced across diverse datasets and attack scenarios. Using zero-shot prompts and image uploads, the study demonstrates GPT-4's ability to extract iris features, perform soft biometric reasoning, detect presentation attacks, and engage in cross-modality matching, often outperforming Gemini in depth and usability. Key contributions include a detailed prompt-engineering approach, robust iris analysis under challenging conditions, and a nuanced comparison with a contemporary AI model, highlighting both capabilities and limitations. The work suggests a promising path for integrating LLMs into adaptive, interactive biometric security systems and calls for further cross-LLM evaluations and design optimizations to maximize real-world robustness and explainability.

Abstract

This study utilizes the advanced capabilities of the GPT-4 multimodal Large Language Model (LLM) to explore its potential in iris recognition - a field less common and more specialized than face recognition. By focusing on this niche yet crucial area, we investigate how well AI tools like ChatGPT can understand and analyze iris images. Through a series of meticulously designed experiments employing a zero-shot learning approach, the capabilities of ChatGPT-4 was assessed across various challenging conditions including diverse datasets, presentation attacks, occlusions such as glasses, and other real-world variations. The findings convey ChatGPT-4's remarkable adaptability and precision, revealing its proficiency in identifying distinctive iris features, while also detecting subtle effects like makeup on iris recognition. A comparative analysis with Gemini Advanced - Google's AI model - highlighted ChatGPT-4's better performance and user experience in complex iris analysis tasks. This research not only validates the use of LLMs for specialized biometric applications but also emphasizes the importance of nuanced query framing and interaction design in extracting significant insights from biometric data. Our findings suggest a promising path for future research and the development of more adaptable, efficient, robust and interactive biometric security solutions.

ChatGPT Meets Iris Biometrics

TL;DR

The paper investigates whether a multimodal LLM, GPT-4, can analyze iris images for biometric recognition and compares its performance to Google's Gemini Advanced across diverse datasets and attack scenarios. Using zero-shot prompts and image uploads, the study demonstrates GPT-4's ability to extract iris features, perform soft biometric reasoning, detect presentation attacks, and engage in cross-modality matching, often outperforming Gemini in depth and usability. Key contributions include a detailed prompt-engineering approach, robust iris analysis under challenging conditions, and a nuanced comparison with a contemporary AI model, highlighting both capabilities and limitations. The work suggests a promising path for integrating LLMs into adaptive, interactive biometric security systems and calls for further cross-LLM evaluations and design optimizations to maximize real-world robustness and explainability.

Abstract

This study utilizes the advanced capabilities of the GPT-4 multimodal Large Language Model (LLM) to explore its potential in iris recognition - a field less common and more specialized than face recognition. By focusing on this niche yet crucial area, we investigate how well AI tools like ChatGPT can understand and analyze iris images. Through a series of meticulously designed experiments employing a zero-shot learning approach, the capabilities of ChatGPT-4 was assessed across various challenging conditions including diverse datasets, presentation attacks, occlusions such as glasses, and other real-world variations. The findings convey ChatGPT-4's remarkable adaptability and precision, revealing its proficiency in identifying distinctive iris features, while also detecting subtle effects like makeup on iris recognition. A comparative analysis with Gemini Advanced - Google's AI model - highlighted ChatGPT-4's better performance and user experience in complex iris analysis tasks. This research not only validates the use of LLMs for specialized biometric applications but also emphasizes the importance of nuanced query framing and interaction design in extracting significant insights from biometric data. Our findings suggest a promising path for future research and the development of more adaptable, efficient, robust and interactive biometric security solutions.
Paper Structure (15 sections, 13 figures, 3 tables)

This paper contains 15 sections, 13 figures, 3 tables.

Table of Contents

  1. Introduction
  2. Background
  3. Our Approach
  4. ChatGPT-4
  5. Gemini
  6. Experimental Results
  7. Soft Biometrics
  8. One-to-one Verification
  9. Similarity Scores
  10. ChatGPT vs. VeriEye
  11. PA Detection
  12. Multiple Images
  13. Cross-Modality Match
  14. Gemini Results
  15. Summary and Future Work

Figures (13)

  • Figure 1: Example of the interaction for iris verification. The figure illustrates a one-to-one matching process where specific prompts are provided to the ChatGPT, followed by its responses. This interaction showcases the GPT-4's ability to process and verify iris data through conversational prompts, highlighting its potential utility in biometric authentication systems.
  • Figure 2: ChatGPT was asked to return a similarity score (0-100) and a binary decision ('yes' or 'no') for each iris verification prompt.
  • Figure 3: CASIA-Genuine
  • Figure 4: IITD-Genuine
  • Figure 5: ND-Genuine
  • ...and 8 more figures