Deep Shape from Polarization

TL;DR

This paper introduces a deep learning approach for Shape from Polarization that integrates physical models as priors, outperforming previous methods on a new diverse dataset.

Contribution

It is the first to combine physics-based priors with deep learning for SfP, achieving superior accuracy across various conditions.

Findings

Achieved lowest test error on all tested conditions

Outperformed previous state-of-the-art methods

Validated the effectiveness of blending physics priors with neural networks

Abstract

This paper makes a first attempt to bring the Shape from Polarization (SfP) problem to the realm of deep learning. The previous state-of-the-art methods for SfP have been purely physics-based. We see value in these principled models, and blend these physical models as priors into a neural network architecture. This proposed approach achieves results that exceed the previous state-of-the-art on a challenging dataset we introduce. This dataset consists of polarization images taken over a range of object textures, paints, and lighting conditions. We report that our proposed method achieves the lowest test error on each tested condition in our dataset, showing the value of blending data-driven and physics-driven approaches.