Logo
ScienceStack
Table of Contents
Fetching ...
Sign In

VLAgeBench: Benchmarking Large Vision-Language Models for Zero-Shot Human Age Estimation

Rakib Hossain Sajib, Md Kishor Morol, Rajan Das Gupta, Mohammad Sakib Mahmood, Shuvra Smaran Das

Abstract

Human age estimation from facial images represents a challenging computer vision task with significant applications in biometrics, healthcare, and human-computer interaction. While traditional deep learning approaches require extensive labeled datasets and domain-specific training, recent advances in large vision-language models (LVLMs) offer the potential for zero-shot age estimation. This study presents a comprehensive zero-shot evaluation of state-of-the-art Large Vision-Language Models (LVLMs) for facial age estimation, a task traditionally dominated by domain-specific convolutional networks and supervised learning. We assess the performance of GPT-4o, Claude 3.5 Sonnet, and LLaMA 3.2 Vision on two benchmark datasets, UTKFace and FG-NET, without any fine-tuning or task-specific adaptation. Using eight evaluation metrics, including MAE, MSE, RMSE, MAPE, MBE, $R^2$, CCC, and $\pm$5-year accuracy, we demonstrate that general-purpose LVLMs can deliver competitive performance in zero-shot settings. Our findings highlight the emergent capabilities of LVLMs for accurate biometric age estimation and position these models as promising tools for real-world applications. Additionally, we highlight performance disparities linked to image quality and demographic subgroups, underscoring the need for fairness-aware multimodal inference. This work introduces a reproducible benchmark and positions LVLMs as promising tools for real-world applications in forensic science, healthcare monitoring, and human-computer interaction. The benchmark focuses on strict zero-shot inference without fine-tuning and highlights remaining challenges related to prompt sensitivity, interpretability, computational cost, and demographic fairness.

VLAgeBench: Benchmarking Large Vision-Language Models for Zero-Shot Human Age Estimation

Abstract

Human age estimation from facial images represents a challenging computer vision task with significant applications in biometrics, healthcare, and human-computer interaction. While traditional deep learning approaches require extensive labeled datasets and domain-specific training, recent advances in large vision-language models (LVLMs) offer the potential for zero-shot age estimation. This study presents a comprehensive zero-shot evaluation of state-of-the-art Large Vision-Language Models (LVLMs) for facial age estimation, a task traditionally dominated by domain-specific convolutional networks and supervised learning. We assess the performance of GPT-4o, Claude 3.5 Sonnet, and LLaMA 3.2 Vision on two benchmark datasets, UTKFace and FG-NET, without any fine-tuning or task-specific adaptation. Using eight evaluation metrics, including MAE, MSE, RMSE, MAPE, MBE, , CCC, and 5-year accuracy, we demonstrate that general-purpose LVLMs can deliver competitive performance in zero-shot settings. Our findings highlight the emergent capabilities of LVLMs for accurate biometric age estimation and position these models as promising tools for real-world applications. Additionally, we highlight performance disparities linked to image quality and demographic subgroups, underscoring the need for fairness-aware multimodal inference. This work introduces a reproducible benchmark and positions LVLMs as promising tools for real-world applications in forensic science, healthcare monitoring, and human-computer interaction. The benchmark focuses on strict zero-shot inference without fine-tuning and highlights remaining challenges related to prompt sensitivity, interpretability, computational cost, and demographic fairness.

Paper Structure

This paper contains 20 sections, 8 equations, 6 figures, 1 table, 1 algorithm.

Table of Contents

  1. Introduction
  2. Literature Review
  3. Methodology
  4. Datasets
  5. Models
  6. Evaluation Metrics
  7. Mean Absolute Error (MAE)
  8. Mean Squared Error (MSE)
  9. Root Mean Squared Error (RMSE)
  10. Mean Absolute Percentage Error (MAPE)
  11. Mean Bias Error (MBE)
  12. Coefficient of Determination ($R^2$)
  13. Concordance Correlation Coefficient (CCC)
  14. Accuracy within ±5 Years
  15. Experimental Setup
  16. ...and 5 more sections

Figures (6)

  • Figure 1: Pipeline for Age Estimation Using Large Vision-Language Models (LVLMs). Input images are processed by three LVLMs (LLaMA 3.2 Vision, GPT-4o, and Claude 3.5 Sonnet) to generate individual age predictions, which are then collected in a unified prediction table (CSV) and evaluated using identical performance metrics.
  • Figure 2: Zero-shot facial age estimation by LVLMs. (a) An input image and standardized prompt instruct the model to return a numeric age without explanation. (b) Predictions from LLaMA 3.2 Vision, GPT-4o, and Claude 3.5 Sonnet are shown, demonstrating their numeric outputs for the same input.
  • Figure 3: Model Performance Comparison on UTKFace Dataset
  • Figure 4: Model Performance Comparison on FG-NET Dataset
  • Figure 5: Radar chart visualization of model performance on UTKFace dataset, illustrating the relative advantages of each LVLM across different evaluation criteria.
  • ...and 1 more figures