CLERF: Contrastive LEaRning for Full Range Head Pose Estimation

TL;DR

This paper presents CLERF, a contrastive learning framework that leverages 3D-aware GANs to improve full-range head pose estimation, especially under pose variations and data sparsity, achieving state-of-the-art results.

Contribution

Introducing a novel contrastive learning approach combined with 3D-aware GANs for accurate, full-range head pose estimation, including upside-down poses, outperforming existing models.

Findings

Performs on par with state-of-the-art on standard datasets.

Outperforms existing models on rotated/flipped images.

First to accurately predict full-range head poses including upside-down.

Abstract

We introduce a novel framework for representation learning in head pose estimation (HPE). Previously such a scheme was difficult due to head pose data sparsity, making triplet sampling infeasible. Recent progress in 3D generative adversarial networks (3D-aware GAN) has opened the door for easily sampling triplets (anchor, positive, negative). We perform contrastive learning on extensively augmented data including geometric transformations and demonstrate that contrastive learning allows networks to learn genuine features that contribute to accurate HPE. On the other hand, we observe that existing HPE works struggle to predict head poses as accurately when test image rotation matrices are slightly out of the training dataset distribution. Experiments show that our methodology performs on par with state-of-the-art models on standard test datasets and outperforms them when images are slightly rotated/ flipped or full range head pose. To the best of our knowledge, we are the first to deliver a true full range HPE model capable of accurately predicting any head pose including upside-down pose. Furthermore, we compared with other existing full-yaw range models and demonstrated superior results.