Papers › ProDepth: Boosting Self-Supervised Multi-Frame Monocular Depth with Probabilistic Fusion

ProDepth: Boosting Self-Supervised Multi-Frame Monocular Depth with Probabilistic Fusion

12 Jul 2024arXiv:2407.09303archive 2025-07-28

Sungmin Woo, Wonjoon Lee, Woo Jin Kim, Dogyoon Lee, Sangyoun Lee

Self-supervised multi-frame monocular depth estimation relies on the geometric consistency between successive frames under the assumption of a static scene. However, the presence of moving objects in dynamic scenes introduces inevitable inconsistencies, causing misaligned multi-frame feature matching and misleading self-supervision during training. In this paper, we propose a novel framework called ProDepth, which effectively addresses the mismatch problem caused by dynamic objects using a probabilistic approach. We initially deduce the uncertainty associated with static scene assumption by adopting an auxiliary decoder. This decoder analyzes inconsistencies embedded in the cost volume, inferring the probability of areas being dynamic. We then directly rectify the erroneous cost volume for dynamic areas through a Probabilistic Cost Volume Modulation (PCVM) module. Specifically, we derive probability distributions of depth candidates from both single-frame and multi-frame cues, modulating the cost volume by adaptively fusing those distributions based on the inferred uncertainty. Additionally, we present a self-supervision loss reweighting strategy that not only masks out incorrect supervision with high uncertainty but also mitigates the risks in remaining possible dynamic areas in accordance with the probability. Our proposed method excels over state-of-the-art approaches in all metrics on both Cityscapes and KITTI datasets, and demonstrates superior generalization ability on the Waymo Open dataset.

PaperPDFCode

In Syntology Open this paper in Syntology's Atlas, the map of the papers in Syntology's graph and their citations.

Code

sungmin-woo/ProDepth officialmentioned on GitHubpytorch report

Repository list and official/mentioned flags are the archive's, frozen 2025-07-28. Reachability, where shown, is from one Syntology probe window (2026-09-16 to 2026-09-18); repositories not probed show nothing. GitHub stars are not tracked.

Code Syntology ran Syntology

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

Tasks

DecoderDepth EstimationDepth PredictionMonocular Depth EstimationUnsupervised Monocular Depth Estimation

Results from the paper archive 2025-07-28

TaskDatasetModelMetricValueRank at snapshotLeaderboardReport
Monocular Depth Estimation KITTI Eigen split unsupervised ProDepth Delta < 1.25 0.918 #3 of 55 Archive leaderboard report
Monocular Depth Estimation KITTI Eigen split unsupervised ProDepth Delta < 1.25^2 0.969 #3 of 55 Archive leaderboard report
Monocular Depth Estimation KITTI Eigen split unsupervised ProDepth Delta < 1.25^3 0.984 #3 of 55 Archive leaderboard report
Monocular Depth Estimation KITTI Eigen split unsupervised ProDepth Mono O #3 of 55 Archive leaderboard report
Monocular Depth Estimation KITTI Eigen split unsupervised ProDepth RMSE 4.139 #3 of 55 Archive leaderboard report
Monocular Depth Estimation KITTI Eigen split unsupervised ProDepth RMSE log 0.166 #3 of 55 Archive leaderboard report
Monocular Depth Estimation KITTI Eigen split unsupervised ProDepth Resolution 640x192 #3 of 55 Archive leaderboard report
Monocular Depth Estimation KITTI Eigen split unsupervised ProDepth Sq Rel 0.629 #3 of 55 Archive leaderboard report
Monocular Depth Estimation KITTI Eigen split unsupervised ProDepth Test frames 2(-1,0) #3 of 55 Archive leaderboard report
Monocular Depth Estimation KITTI Eigen split unsupervised ProDepth absolute relative error 0.086 #3 of 55 Archive leaderboard report
Monocular Depth Estimation KITTI Eigen split unsupervised ProDepth(M+640x192) Delta < 1.25 0.902 #16 of 55 Archive leaderboard report
Monocular Depth Estimation KITTI Eigen split unsupervised ProDepth(M+640x192) Delta < 1.25^2 0.967 #16 of 55 Archive leaderboard report
Monocular Depth Estimation KITTI Eigen split unsupervised ProDepth(M+640x192) Delta < 1.25^3 0.985 #16 of 55 Archive leaderboard report
Monocular Depth Estimation KITTI Eigen split unsupervised ProDepth(M+640x192) Mono O #16 of 55 Archive leaderboard report
Monocular Depth Estimation KITTI Eigen split unsupervised ProDepth(M+640x192) RMSE 4.345 #16 of 55 Archive leaderboard report
Monocular Depth Estimation KITTI Eigen split unsupervised ProDepth(M+640x192) RMSE log 0.172 #16 of 55 Archive leaderboard report
Monocular Depth Estimation KITTI Eigen split unsupervised ProDepth(M+640x192) Resolution 640x192 #16 of 55 Archive leaderboard report
Monocular Depth Estimation KITTI Eigen split unsupervised ProDepth(M+640x192) Sq Rel 0.693 #16 of 55 Archive leaderboard report
Monocular Depth Estimation KITTI Eigen split unsupervised ProDepth(M+640x192) absolute relative error 0.095 #16 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