MarrNet: 3D Shape Reconstruction via 2.5D Sketches

TL;DR

MarrNet introduces a sequential, end-to-end trainable model that reconstructs 3D shapes from single images by estimating 2.5D sketches, effectively addressing domain adaptation issues and achieving state-of-the-art results.

Contribution

The paper presents a novel two-step approach that estimates 2.5D sketches before 3D reconstruction, enabling training on synthetic data and improving real-world applicability.

Findings

Achieves state-of-the-art 3D reconstruction accuracy.

Effectively transfers from synthetic to real data.

No human annotations needed for training.

Abstract

3D object reconstruction from a single image is a highly under-determined problem, requiring strong prior knowledge of plausible 3D shapes. This introduces challenges for learning-based approaches, as 3D object annotations are scarce in real images. Previous work chose to train on synthetic data with ground truth 3D information, but suffered from domain adaptation when tested on real data. In this work, we propose MarrNet, an end-to-end trainable model that sequentially estimates 2.5D sketches and 3D object shape. Our disentangled, two-step formulation has three advantages. First, compared to full 3D shape, 2.5D sketches are much easier to be recovered from a 2D image; models that recover 2.5D sketches are also more likely to transfer from synthetic to real data. Second, for 3D reconstruction from 2.5D sketches, systems can learn purely from synthetic data. This is because we can easily render realistic 2.5D sketches without modeling object appearance variations in real images, including lighting, texture, etc. This further relieves the domain adaptation problem. Third, we derive differentiable projective functions from 3D shape to 2.5D sketches; the framework is therefore end-to-end trainable on real images, requiring no human annotations. Our model achieves state-of-the-art performance on 3D shape reconstruction.