AI Paper Atlas

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

Pose-Aware Self-Supervised Learning with Viewpoint Trajectory Regularization

Jiayun Wang, Yubei Chen, Stella X. Yu

ECCV 2024 · Label-efficient and robust learning › Self-supervised and contrastive · approach: Invariance-based self-supervised learning · 5 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 trajectory regularization loss to VICReg (ablation)VICReg ResNet-18 · ShapeNet in-domain relative poserelative pose accuracy (%): 76.7 -> 80.5Improved
add trajectory regularization loss to VICReg (ablation)VICReg ResNet-18 · ShapeNet in-domain semantic classificationaccuracy (%): 85.6 -> 85.6No clear effect
add trajectory regularization loss to VICReg (ablation)VICReg ResNet-18 · ShapeNet unseen posesrelative pose accuracy (%): 73.1 -> 77.5Improved
add trajectory regularization loss to VICReg (ablation)VICReg ResNet-18 · Carvana real photospose accuracy (%): 88.7 -> 91.7Improved
add trajectory regularization loss to SimCLR (ablation)SimCLR ResNet-18 · ShapeNet in-domain absolute poseabsolute pose accuracy (%): 84.8 -> 86.4Improved
use mid-layer conv3 representation instead of last feature layer for pose probing (ablation)VICReg+traj. · ShapeNet unseen posesrelative pose accuracy (%): 62.7 -> 82.6Improved
use mid-layer conv3 representation instead of feature layer (ablation)VICReg+traj. · ShapeNet in-domain relative poserelative pose accuracy (%): 80.5 -> 89.4Improved
use conv3 or conv4 instead of feature layer for semantic classification (ablation)SSL ResNet-18 · ShapeNet semantic classificationaccuracy: <1%No clear effect
compress conv3 representation to 512 dims (ablation)VICReg+traj. conv3 · ShapeNet relative poserelative pose accuracy (%): 87.8 -> 82.4Hurt
compress conv3 representation to 512 dims (ablation)VICReg+traj. conv3 · ShapeNet absolute poseabsolute pose accuracy (%): 92.5 -> 91.4No clear effect
use non-equidistant adjacent poses in triplets (ablation)VICReg+traj. · ShapeNet in-domain relative poserelative pose accuracy (%): 80.5 -> 80.3No clear effect
change layer where trajectory loss is applied (ablation)VICReg+traj. · ShapeNet relative poserelative pose accuracy: <2%No clear effect

Earlier methods this paper compares with

Named baselines from earlier accepted papers in this paper's results tables: better than 1 of 1 on most shared results.

Matched by method name in results tables read by AI. Authors' own tables tend to favour the newer method, and a comparison can be missed when a method is renamed.

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