AI Paper Atlas

Research map / 3D vision / Face and shape

Online Adaptation for Consistent Mesh Reconstruction in the Wild

Xueting Li, Sifei Liu, Shalini De Mello, Kihwan Kim, Xiaolong Wang, Ming-Hsuan Yang, Jan Kautz

NeurIPS 2020 · 3D vision › Face and shape · approach: Articulated and non-rigid shape learning · 18 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 motion deformation branch (asymmetric dV) (ablation)ACMR · CUB test (image)Mask IoU: 0.647 -> 0.708Improved
add motion deformation branch (asymmetric dV) (ablation)ACMR · CUB test (image)PCK@0.1: 0.790 -> 0.855Improved
add ARAP constraint in image training (ablation)ACMR · CUB test (image)Mask IoU: 0.758 -> 0.708Hurt
add ARAP constraint in image training (ablation)ACMR · CUB test (image)PCK@0.1: 0.857 -> 0.855No clear effect
use 8 shape bases instead of a single template (ablation)ACMR · CUB test (image)Mask IoU: 0.605 -> 0.708Improved
use 8 shape bases instead of a single template (ablation)ACMR · CUB test (image)PCK@0.1: 0.655 -> 0.855Improved
asymmetric shape with bases and ARAP (ACMR) vs CMRCMR baseline · CUB test (image)PCK@0.1: 0.810 -> 0.855Improved
switch encoder to ResNet50 with group norm (ACMR-vid)ACMR · CUB test (image)Mask IoU: 0.708 -> 0.773Improved
test-time tuning ResNet18 BN model on video (ablation)ACMR · Bird video datasetPCK@0.1: 0.751 -> 0.424Hurt
online adaptation on test video (no invariance losses) (ablation)ACMR-vid · Bird video datasetJ mean: 0.686 -> 0.836Improved
add part, texture, base shape invariance losses in online adaptation (ablation)ACMR-vid (T) · Bird video datasetF mean: 0.666 -> 0.756Improved
add part, texture, base shape invariance losses in online adaptation (ablation)ACMR-vid (T) · Bird video datasetPCK@0.1: 0.644 -> 0.794Improved

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