Neural Architecture Search for Efficient Uncalibrated Deep Photometric Stereo

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.

Self-calibrating Photometric Stereo by Neural Inverse Rendering

“…Later works [8,11] further improved this pipeline by predicting both the light directions and surface normal at the same time. A recent work [40] proposes a way to search for the most efficient neural architecture for uncalibrated photometric stereo. These neural network methods learn prior information for solving the GBR ambiguity from a large amount of training data with ground truth.…”