Sparse-to-Dense: Depth Prediction from Sparse Depth Samples and a Single Image

TL;DR

This paper presents a deep learning approach that combines sparse depth samples with a single RGB image to improve dense depth prediction accuracy, demonstrating significant error reduction and reliability improvements on indoor and outdoor datasets.

Contribution

The paper introduces a deep regression network that effectively fuses sparse depth samples with RGB images for enhanced depth prediction, including applications in SLAM and LiDAR super-resolution.

Findings

Adding 100 sparse depth samples reduces RMSE by 50% on NYU-Depth-v2.

Increases reliable prediction from 59% to 92% on KITTI.

Enables dense map generation and LiDAR super-resolution.

Abstract

We consider the problem of dense depth prediction from a sparse set of depth measurements and a single RGB image. Since depth estimation from monocular images alone is inherently ambiguous and unreliable, to attain a higher level of robustness and accuracy, we introduce additional sparse depth samples, which are either acquired with a low-resolution depth sensor or computed via visual Simultaneous Localization and Mapping (SLAM) algorithms. We propose the use of a single deep regression network to learn directly from the RGB-D raw data, and explore the impact of number of depth samples on prediction accuracy. Our experiments show that, compared to using only RGB images, the addition of 100 spatially random depth samples reduces the prediction root-mean-square error by 50% on the NYU-Depth-v2 indoor dataset. It also boosts the percentage of reliable prediction from 59% to 92% on the KITTI dataset. We demonstrate two applications of the proposed algorithm: a plug-in module in SLAM to convert sparse maps to dense maps, and super-resolution for LiDARs. Software and video demonstration are publicly available.