AI Paper Atlas

Research map / 3D vision / Depth and stereo

Adaptive Co-Teaching for Unsupervised Monocular Depth Estimation

Weisong Ren, Lijun Wang, Yongri Piao, Miao Zhang, Huchuan Lu, Ting Liu

ECCV 2022 · 3D vision › Depth and stereo · approach: Dense depth prediction and completion · 25 citations (OpenAlex)

Read the paperCheck your idea against 137,217 papers

AI-extracted. The measured effects below were read by AI from the paper's full text (no summary yet). Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.

Measured effects reported in the paper

What was changedWhereMeasuredEffect
use 16-channel depth basis decoder (averaged) (ablation)Baseline · KITTI Eigen (192x640)AbsRel: 0.113 -> 0.112No clear effect
learn coefficients on top of depth bases (MUSTNet) (ablation)+BD · KITTI Eigen (192x640)AbsRel: 0.112 -> 0.109Improved
use MUSTNet teacher instead of Baseline teacher (ablation)Baseline student · KITTI Eigen (192x640)AbsRel: 0.113 -> 0.112Improved
use MUSTNet student architecture instead of Baseline (ablation)student with MUSTNet teacher · KITTI Eigen (192x640)AbsRel: 0.112 -> 0.110Improved
distill from single teacher (ablation)MUSTNet · KITTI Eigen (192x640)AbsRel: 0.109 -> 0.110No clear effect
distill from bootstrapped 4-teacher ensemble (4x parameters) (ablation)MUSTNet · KITTI Eigen (192x640)AbsRel: 0.110 -> 0.108Improved
distill from compact MUSTNet teacher ensemble with photometric and pseudo-label losses (ablation)MUSTNet · KITTI Eigen (192x640)AbsRel: 0.110 -> 0.108Improved
add selective masks (cost-aware co-teaching loss) (ablation)adaptive co-teaching · KITTI Eigen (192x640)AbsRel: 0.108 -> 0.106Improved
add adaptive co-teaching with dual-stream decoderMD2 (Monodepth2) · KITTI Eigen (192x640)AbsRel: 0.115 -> 0.108Improved
use dual-stream decoder alone (ablation)MD2 · KITTI Eigen (192x640)AbsRel: 0.115 -> 0.111Improved
use proposed method instead of HR-DepthHR-Depth · nuScenes-miniAbsRel: 0.196 -> 0.192Improved

Closest papers by meaning

Writing a paper in this area? scime draft is a free LaTeX editor with AI included, real-time co-authors and a check of your references and venue rules before you submit. It also imports OpenAI Prism projects.

Open scime draft

Get the next scime Dispatch and a monthly field update

Free. We send a confirmation link first; every email has a one-click unsubscribe. Privacy