Improving Facial Landmark Detection Accuracy and Efficiency with Knowledge Distillation

TL;DR

This paper presents a knowledge distillation approach to develop lightweight, accurate, and efficient facial landmark detection models suitable for embedded systems, addressing challenges of diversity, robustness, and resource constraints.

Contribution

It introduces a novel knowledge distillation method specifically designed for facial landmark detection, improving model efficiency without sacrificing accuracy.

Findings

Achieved top 6th place in IEEE ICME 2024 PAIR competition

Developed models capable of real-time performance on embedded devices

Enhanced robustness across diverse facial expressions and conditions

Abstract

The domain of computer vision has experienced significant advancements in facial-landmark detection, becoming increasingly essential across various applications such as augmented reality, facial recognition, and emotion analysis. Unlike object detection or semantic segmentation, which focus on identifying objects and outlining boundaries, faciallandmark detection aims to precisely locate and track critical facial features. However, deploying deep learning-based facial-landmark detection models on embedded systems with limited computational resources poses challenges due to the complexity of facial features, especially in dynamic settings. Additionally, ensuring robustness across diverse ethnicities and expressions presents further obstacles. Existing datasets often lack comprehensive representation of facial nuances, particularly within populations like those in Taiwan. This paper introduces a novel approach to address these challenges through the development of a knowledge distillation method. By transferring knowledge from larger models to smaller ones, we aim to create lightweight yet powerful deep learning models tailored specifically for facial-landmark detection tasks. Our goal is to design models capable of accurately locating facial landmarks under varying conditions, including diverse expressions, orientations, and lighting environments. The ultimate objective is to achieve high accuracy and real-time performance suitable for deployment on embedded systems. This method was successfully implemented and achieved a top 6th place finish out of 165 participants in the IEEE ICME 2024 PAIR competition.