Shape from Shading through Shape Evolution

TL;DR

This paper introduces a novel shape-from-shading method that uses an iterative process of shape evolution and deep network training with synthetic images, eliminating the need for external shape datasets.

Contribution

It proposes a unique approach combining shape evolution and deep learning, improving shape-from-shading without relying on external shape datasets.

Findings

Achieves state-of-the-art performance on benchmark datasets.

Demonstrates effective shape evolution guided by network training.

Eliminates dependency on external shape datasets for synthetic image generation.

Abstract

In this paper, we address the shape-from-shading problem by training deep networks with synthetic images. Unlike conventional approaches that combine deep learning and synthetic imagery, we propose an approach that does not need any external shape dataset to render synthetic images. Our approach consists of two synergistic processes: the evolution of complex shapes from simple primitives, and the training of a deep network for shape-from-shading. The evolution generates better shapes guided by the network training, while the training improves by using the evolved shapes. We show that our approach achieves state-of-the-art performance on a shape-from-shading benchmark.