SA-ConvONet: Sign-Agnostic Optimization of Convolutional Occupancy Networks

TL;DR

SA-ConvONet introduces a sign-agnostic optimization method for convolutional occupancy networks, enabling accurate surface reconstruction from raw point clouds without requiring surface normals, thus improving scalability and generality.

Contribution

It extends sign-agnostic learning to local shape modeling with convolutional occupancy networks, allowing unified, scalable, and normal-free surface reconstruction.

Findings

Outperforms previous methods on object and scene datasets

Achieves higher reconstruction accuracy from un-oriented point clouds

Demonstrates robustness without surface normal information

Abstract

Surface reconstruction from point clouds is a fundamental problem in the computer vision and graphics community. Recent state-of-the-arts solve this problem by individually optimizing each local implicit field during inference. Without considering the geometric relationships between local fields, they typically require accurate normals to avoid the sign conflict problem in overlapped regions of local fields, which severely limits their applicability to raw scans where surface normals could be unavailable. Although SAL breaks this limitation via sign-agnostic learning, further works still need to explore how to extend this technique for local shape modeling. To this end, we propose to learn implicit surface reconstruction by sign-agnostic optimization of convolutional occupancy networks, to simultaneously achieve advanced scalability to large-scale scenes, generality to novel shapes, and applicability to raw scans in a unified framework. Concretely, we achieve this goal by a simple yet effective design, which further optimizes the pre-trained occupancy prediction networks with an unsigned cross-entropy loss during inference. The learning of occupancy fields is conditioned on convolutional features from an hourglass network architecture. Extensive experimental comparisons with previous state-of-the-arts on both object-level and scene-level datasets demonstrate the superior accuracy of our approach for surface reconstruction from un-orientated point clouds. The code is available at https://github.com/tangjiapeng/SA-ConvONet.