Logo
ScienceStack
Table of Contents
Fetching ...
Sign In

Deep Learning Approach for Ear Recognition and Longitudinal Evaluation in Children

Afzal Hossain, Tipu Sultan, Stephanie Schuckers

TL;DR

The paper addresses the challenge of ear-based biometrics in children, where rapid ear development impairs longitudinal identification. It introduces a deep learning pipeline that uses Mask R-CNN for ear segmentation and a VGG16-MobileNet ensemble for feature extraction, evaluated on a newly collected longitudinal dataset of children aged 4–14 over 2.5 years, plus an adult IITD baseline. Key findings show strong within-session performance (TAR > 90%, FAR ≈ 2%) but substantial drop in cross-session accuracy over time (55–76% TAR across 30 months), with especially low performance for sub-8-year-olds due to rapid growth. The study highlights the need for adaptive, time-aware recognition strategies and potentially multi-modal approaches to robustly identify children across developmental stages.

Abstract

Ear recognition as a biometric modality is becoming increasingly popular, with promising broader application areas. While current applications involve adults, one of the challenges in ear recognition for children is the rapid structural changes in the ear as they age. This work introduces a foundational longitudinal dataset collected from children aged 4 to 14 years over a 2.5-year period and evaluates ear recognition performance in this demographic. We present a deep learning based approach for ear recognition, using an ensemble of VGG16 and MobileNet, focusing on both adult and child datasets, with an emphasis on longitudinal evaluation for children.

Deep Learning Approach for Ear Recognition and Longitudinal Evaluation in Children

TL;DR

The paper addresses the challenge of ear-based biometrics in children, where rapid ear development impairs longitudinal identification. It introduces a deep learning pipeline that uses Mask R-CNN for ear segmentation and a VGG16-MobileNet ensemble for feature extraction, evaluated on a newly collected longitudinal dataset of children aged 4–14 over 2.5 years, plus an adult IITD baseline. Key findings show strong within-session performance (TAR > 90%, FAR ≈ 2%) but substantial drop in cross-session accuracy over time (55–76% TAR across 30 months), with especially low performance for sub-8-year-olds due to rapid growth. The study highlights the need for adaptive, time-aware recognition strategies and potentially multi-modal approaches to robustly identify children across developmental stages.

Abstract

Ear recognition as a biometric modality is becoming increasingly popular, with promising broader application areas. While current applications involve adults, one of the challenges in ear recognition for children is the rapid structural changes in the ear as they age. This work introduces a foundational longitudinal dataset collected from children aged 4 to 14 years over a 2.5-year period and evaluates ear recognition performance in this demographic. We present a deep learning based approach for ear recognition, using an ensemble of VGG16 and MobileNet, focusing on both adult and child datasets, with an emphasis on longitudinal evaluation for children.
Paper Structure (8 sections, 7 figures, 1 table)

This paper contains 8 sections, 7 figures, 1 table.

Table of Contents

  1. Introduction
  2. Related Works
  3. Network Architecture
  4. Experiment
  5. Dataset and Pre-processing of Data
  6. Training and Evaluation Procedure
  7. Results and Discussions
  8. Conclusion

Figures (7)

  • Figure 1: The Mask R-CNN framework for instance segmentation
  • Figure 2: Example images from zoomed profile face and corresponding processed ear images
  • Figure 3: Performance Comparison of TAR and FAR Across Different Methods for IITD dataset
  • Figure 4: Performance evaluation of our ensemble method (combination of VGG16 and MobileNet) across six collections (Col-1 to Col-6) of our child dataset
  • Figure 5: TAR% at 2.0% FAR for increasing time gaps between enrollment and verification
  • ...and 2 more figures