Neural Architecture Search for Efficient Uncalibrated Deep Photometric Stereo

TL;DR

This paper introduces a neural architecture search method for uncalibrated photometric stereo, automatically designing lightweight neural networks that achieve high surface normal accuracy with lower memory requirements.

Contribution

It proposes a differentiable NAS approach tailored for uncalibrated PS, explicitly handling task-specific constraints and GBR ambiguity, improving over handcrafted architectures.

Findings

Achieves state-of-the-art performance on DiLiGenT dataset.

Reduces memory footprint compared to existing methods.

Automatically discovers efficient neural architectures for uncalibrated PS.

Abstract

We present an automated machine learning approach for uncalibrated photometric stereo (PS). Our work aims at discovering lightweight and computationally efficient PS neural networks with excellent surface normal accuracy. Unlike previous uncalibrated deep PS networks, which are handcrafted and carefully tuned, we leverage differentiable neural architecture search (NAS) strategy to find uncalibrated PS architecture automatically. We begin by defining a discrete search space for a light calibration network and a normal estimation network, respectively. We then perform a continuous relaxation of this search space and present a gradient-based optimization strategy to find an efficient light calibration and normal estimation network. Directly applying the NAS methodology to uncalibrated PS is not straightforward as certain task-specific constraints must be satisfied, which we impose explicitly. Moreover, we search for and train the two networks separately to account for the Generalized Bas-Relief (GBR) ambiguity. Extensive experiments on the DiLiGenT dataset show that the automatically searched neural architectures performance compares favorably with the state-of-the-art uncalibrated PS methods while having a lower memory footprint.