# Revisiting Depth Completion from a Stereo Matching Perspective for Cross-domain Generalization

Source: [Revisiting Depth Completion from a Stereo Matching Perspective for Cross-domain Generalization](<https://www.sota2.com/research/paper/68fa71731e209131cea51520>)

Collection index: [llms.txt](<https://www.sota2.com/research/llms.txt>)

Data updated: 2025-10-26T02:06:39.354Z

Authors: Luca Bartolomei, Matteo Poggi, Andrea Conti, Fabio Tosi, Stefano Mattoccia

Published: 2023-12-14T18:59:58.000Z

## Abstract

This paper proposes a new framework for depth completion robust against domain-shifting issues. It exploits the generalization capability of modern stereo networks to face depth completion, by processing fictitious stereo pairs obtained through a virtual pattern projection paradigm. Any stereo network or traditional stereo matcher can be seamlessly plugged into our framework, allowing for the deployment of a virtual stereo setup that is future-proof against advancement in the stereo field. Exhaustive experiments on cross-domain generalization support our claims. Hence, we argue that our framework can help depth completion to reach new deployment scenarios.

## Recorded benchmark results

A research result applies to its stated dataset, split, metric and evaluation configuration. Compare like-for-like settings, preserve the metric's unit and direction, and distinguish paper-reported results from independent reproduction. The best result recorded here is not a claim of exhaustive, current global coverage.

- [Depth Completion on DDAD](<https://www.sota2.com/research/sota/depth-completion-on-ddad.md>): Depth Completion; DDAD; MAE: 1.344.

- [Depth Completion on VOID](<https://www.sota2.com/research/sota/depth-completion-on-void.md>): Depth Completion; VOID; MAE: 0.148.

- [Depth Completion on KITTI](<https://www.sota2.com/research/sota/depth-completion-on-kitti.md>): Depth Completion; KITTI; RMSE: 1.609.

- [Depth Completion on Overall Average (ScanNet, IBims-1, VOID, NYUv2, KITTI, DDAD)](<https://www.sota2.com/research/sota/depth-completion-on-overall-average-scannet-ibims-1-void-nyuv2-kitti-ddad.md>): Depth Completion; Overall Average (ScanNet, IBims-1, VOID, NYUv2, KITTI, DDAD); Rank: 4.

- [Depth Completion on ScanNet](<https://www.sota2.com/research/sota/depth-completion-on-scannet.md>): Depth Completion; ScanNet; MAE: 0.023.

- [Depth Completion on IBims-1](<https://www.sota2.com/research/sota/depth-completion-on-ibims-1.md>): Depth Completion; iBIMS-1; MAE: 0.062.

- [Depth Completion on NYU v2](<https://www.sota2.com/research/sota/depth-completion-on-nyu-v2.md>): Depth Completion; NYU V2; RMSE: 0.247.

- [Depth Completion on KITTI (val)](<https://www.sota2.com/research/sota/depth-completion-on-kitti-val.md>): Depth Completion; KITTI (val); Runtime (ms): not recorded.

## Original sources and implementation

[Original paper](<https://arxiv.org/abs/2312.09254>)

[PDF](<https://arxiv.org/pdf/2312.09254.pdf>)