AI Paper Atlas

Research map / Label-efficient and robust learning / Self-supervised and contrastive

Improving Equivariance in State-of-the-Art Supervised Depth and Normal Predictors

Yuanyi Zhong, Anand Bhattad, Yu-Xiong Wang, David Forsyth

ICCV 2023 · Label-efficient and robust learning › Self-supervised and contrastive · approach: Pretext-task visual pre-training · 1 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
add equivariant loss on top of augmentationU-Net · Taskonomy Depth-ZBuffer (supervised, U-Net)delta>1.25 (%): 27.0 -> 25.0Improved
add equivariant loss on top of augmentationU-Net · Taskonomy Edge2DL1 error x10^-3: 7.35 -> 6.35Improved
add equivariant loss on top of augmentationU-Net · Taskonomy Surface Normalangular error (deg): 6.55 -> 6.47Improved
add unlabeled data with equivariant loss (ablation)U-Net, 1/4 labeled · Taskonomy Depth-ZBufferdelta>1.25 (%): 42.0 -> 41.0Improved
add unlabeled data with equivariant lossU-Net, all labeled · Taskonomy Depth-ZBufferdelta>1.25 (%): 25.0 -> 25.0No clear effect
unsupervised equivariant finetuningMiDaS-2.1 CNN · NYU-v2delta>1.25 (%): 8.71 -> 7.82Improved
unsupervised equivariant finetuningMiDaS-3.0 DPT · NYU-v2delta>1.25 (%): 8.32 -> 7.75Improved
unsupervised equivariant finetuningBae et al. normal · NYU-v2mean angular error (deg): 16.03 -> 15.71Improved
add inference-time 3-crop averaging (ablation)MiDaS-v2.1 pre-trained · NYU-v2delta>1.25 (%): 8.71 -> 8.63No clear effect
use feature space (L-1) instead of label space (L) (ablation)EqLoss · Taskonomy Depth-ZBuffer (supervised, U-Net)delta>1.25 (%): 25.7 -> 25.0Improved
apply loss at earlier upsampling block up3 instead of L-1 (ablation)EqLoss · Taskonomy Depth-ZBuffer (supervised, U-Net)delta>1.25 (%): 25.0 -> 26.1Hurt
raise loss coefficient from 1e-4 to 3e-4 (ablation)EqLoss · Taskonomy Depth-ZBuffer (supervised, U-Net)delta>1.25 (%): 25.0 -> 25.8Hurt

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