Papers › DCPI-Depth: Explicitly Infusing Dense Correspondence Prior to Unsupervised Monocular...

DCPI-Depth: Explicitly Infusing Dense Correspondence Prior to Unsupervised Monocular Depth Estimation

27 May 2024arXiv:2405.16960archive 2025-07-28

Mengtan Zhang, Yi Feng, Qijun Chen, Rui Fan

There has been a recent surge of interest in learning to perceive depth from monocular videos in an unsupervised fashion. A key challenge in this field is achieving robust and accurate depth estimation in challenging scenarios, particularly in regions with weak textures or where dynamic objects are present. This study makes three major contributions by delving deeply into dense correspondence priors to provide existing frameworks with explicit geometric constraints. The first novelty is a contextual-geometric depth consistency loss, which employs depth maps triangulated from dense correspondences based on estimated ego-motion to guide the learning of depth perception from contextual information, since explicitly triangulated depth maps capture accurate relative distances among pixels. The second novelty arises from the observation that there exists an explicit, deducible relationship between optical flow divergence and depth gradient. A differential property correlation loss is, therefore, designed to refine depth estimation with a specific emphasis on local variations. The third novelty is a bidirectional stream co-adjustment strategy that enhances the interaction between rigid and optical flows, encouraging the former towards more accurate correspondence and making the latter more adaptable across various scenarios under the static scene hypotheses. DCPI-Depth, a framework that incorporates all these innovative components and couples two bidirectional and collaborative streams, achieves state-of-the-art performance and generalizability across multiple public datasets, outperforming all existing prior arts. Specifically, it demonstrates accurate depth estimation in texture-less and dynamic regions, and shows more reasonable smoothness. Our source code will be publicly available at mias.group/DCPI-Depth upon publication.

PaperPDF

Code

No code repository is listed for this paper in the archive or in Syntology's graph.

Code Syntology ran Syntology

Not run by Syntology. Nothing on this page verifies that the listed code works.

Tasks

Depth EstimationMonocular Depth EstimationOptical Flow EstimationUnsupervised Monocular Depth Estimation

Results from the paper archive 2025-07-28

TaskDatasetModelMetricValueRank at snapshotLeaderboardReport
Monocular Depth Estimation KITTI Eigen split unsupervised DCPI-Depth (M+1024x320) Delta < 1.25 0.914 #6 of 55 Archive leaderboard report
Monocular Depth Estimation KITTI Eigen split unsupervised DCPI-Depth (M+1024x320) Delta < 1.25^2 0.969 #6 of 55 Archive leaderboard report
Monocular Depth Estimation KITTI Eigen split unsupervised DCPI-Depth (M+1024x320) Delta < 1.25^3 0.985 #6 of 55 Archive leaderboard report
Monocular Depth Estimation KITTI Eigen split unsupervised DCPI-Depth (M+1024x320) Mono O #6 of 55 Archive leaderboard report
Monocular Depth Estimation KITTI Eigen split unsupervised DCPI-Depth (M+1024x320) RMSE 4.113 #6 of 55 Archive leaderboard report
Monocular Depth Estimation KITTI Eigen split unsupervised DCPI-Depth (M+1024x320) RMSE log 0.167 #6 of 55 Archive leaderboard report
Monocular Depth Estimation KITTI Eigen split unsupervised DCPI-Depth (M+1024x320) Sq Rel 0.655 #6 of 55 Archive leaderboard report
Monocular Depth Estimation KITTI Eigen split unsupervised DCPI-Depth (M+1024x320) absolute relative error 0.090 #6 of 55 Archive leaderboard report
Monocular Depth Estimation KITTI Eigen split unsupervised DCPI-Depth (M+640x192) Delta < 1.25 0.902 #13 of 55 Archive leaderboard report
Monocular Depth Estimation KITTI Eigen split unsupervised DCPI-Depth (M+640x192) Delta < 1.25^2 0.967 #13 of 55 Archive leaderboard report
Monocular Depth Estimation KITTI Eigen split unsupervised DCPI-Depth (M+640x192) Delta < 1.25^3 0.985 #13 of 55 Archive leaderboard report
Monocular Depth Estimation KITTI Eigen split unsupervised DCPI-Depth (M+640x192) RMSE 4.274 #13 of 55 Archive leaderboard report
Monocular Depth Estimation KITTI Eigen split unsupervised DCPI-Depth (M+640x192) RMSE log 0.170 #13 of 55 Archive leaderboard report
Monocular Depth Estimation KITTI Eigen split unsupervised DCPI-Depth (M+640x192) Sq Rel 0.662 #13 of 55 Archive leaderboard report
Monocular Depth Estimation KITTI Eigen split unsupervised DCPI-Depth (M+640x192) absolute relative error 0.095 #13 of 55 Archive leaderboard report
Monocular Depth Estimation KITTI Eigen split unsupervised DCPI-Depth (M+832x256+SC-V3) RMSE 4.496 #45 of 55 Archive leaderboard report
Monocular Depth Estimation KITTI Eigen split unsupervised DCPI-Depth (M+832x256+SC-V3) Sq Rel 0.679 #45 of 55 Archive leaderboard report
Monocular Depth Estimation KITTI Eigen split unsupervised DCPI-Depth (M+832x256+SC-V3) absolute relative error 0.109 #45 of 55 Archive leaderboard report

Ranks are positions in the archive's leaderboards as they stood at the 2025-07-28 snapshot. Results published since then are not among these rows, so a rank here is not a current standing.

Report a problem or propose a change · a person checks every report against the paper or source before anything changes; decisions are listed on /corrections