AI Paper Atlas

Research map / 3D vision / Face and shape

MVTN: Multi-View Transformation Network for 3D Shape Recognition

Abdullah Hamdi, Silvio Giancola, Bernard Ghanem

ICCV 2021 · 3D vision › Face and shape · approach: Primitive-based shape abstraction · 265 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 MVTN view-point regression (ablation)MVCNN, 12 views · ModelNet40overall accuracy: 90.4 -> 92.6Improved
add MVTN view-point regression (ablation)RotationNet, 12 views · ModelNet40overall accuracy: 91.6 -> 93.2Improved
add MVTN view-point regression (ablation)ViewGCN, 12 views · ModelNet40overall accuracy: 93.0 -> 93.8Improved
add MVTN view-point regression (ablation)ViewGCN, 20 views · ModelNet40overall accuracy: 93.3 -> 93.5Improved
add MVTN view-point regression (ablation)ViewGCN* · ScanObjectNN PB_T50_RSoverall accuracy: 80.5 -> 82.8Improved
regress view-points with MVTN instead of late fusion of point and multi-view networks (ablation)MVCNN with PointNet · ModelNet40overall accuracy: 88.4 -> 91.8Improved
use ResNet-34 backbone instead of ResNet-18 (ablation)MVTN-spherical PointNet · ModelNet40overall accuracy: 93.41 -> 92.83No clear effect
use DGCNN point encoder instead of PointNet (ablation)MVTN-spherical ResNet-18 · ModelNet40overall accuracy: 93.41 -> 93.26No clear effect
use spherical instead of circular initial views (ablation)MVTN ResNet-18 PointNet · ModelNet40overall accuracy: 92.83 -> 93.41Improved
transfer MVTN views learned with MVCNN to ViewGCN (ablation)ViewGCN · ModelNet40overall accuracy: 92.4 -> 93.1Improved
randomize object color in training instead of white (ablation)MVTN · ModelNet40overall accuracy: 93.3 -> 93.4No clear effect
predict view-points directly (MVTN-direct) with more than four views (ablation)MVTN-direct · ModelNet40overall accuracyHurt

Later papers that compared with this method

1 later accepted paper reports MVTN as a baseline in their results tables; 0 of them report a better number on most of the results they share with it.

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