Single-Image Piece-wise Planar 3D Reconstruction via Associative Embedding

TL;DR

This paper introduces a two-stage associative embedding approach for single-image piece-wise planar 3D reconstruction, enabling detection of an arbitrary number of planes with real-time speed, improving flexibility and efficiency over prior fixed-number methods.

Contribution

The proposed method is the first to use associative embedding for flexible, real-time single-image planar 3D reconstruction, overcoming fixed plane number limitations of previous approaches.

Findings

Detects an arbitrary number of planes in a single image.

Operates at 30 frames per second, suitable for real-time applications.

Outperforms existing methods on public datasets.

Abstract

Single-image piece-wise planar 3D reconstruction aims to simultaneously segment plane instances and recover 3D plane parameters from an image. Most recent approaches leverage convolutional neural networks (CNNs) and achieve promising results. However, these methods are limited to detecting a fixed number of planes with certain learned order. To tackle this problem, we propose a novel two-stage method based on associative embedding, inspired by its recent success in instance segmentation. In the first stage, we train a CNN to map each pixel to an embedding space where pixels from the same plane instance have similar embeddings. Then, the plane instances are obtained by grouping the embedding vectors in planar regions via an efficient mean shift clustering algorithm. In the second stage, we estimate the parameter for each plane instance by considering both pixel-level and instance-level consistencies. With the proposed method, we are able to detect an arbitrary number of planes. Extensive experiments on public datasets validate the effectiveness and efficiency of our method. Furthermore, our method runs at 30 fps at the testing time, thus could facilitate many real-time applications such as visual SLAM and human-robot interaction. Code is available at https://github.com/svip-lab/PlanarReconstruction.