AI Paper Atlas

Research map / Probabilistic and scientific modeling / Variational and generative sampling

Probabilistic Orientation Estimation with Matrix Fisher Distributions

David Mohlin, Josephine Sullivan, Gérald Bianchi

NeurIPS 2020 · Probabilistic and scientific modeling › Variational and generative sampling · approach: Gradient estimation for discrete variables · 21 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 a matrix Fisher distribution with negative log-likelihood trainingResNet-101 orientation estimation · Pascal3D+median angle error (deg): 9.20 -> 8.90Improved
Use a matrix Fisher distribution with negative log-likelihood trainingResNet-101 orientation estimation · ModelNet10-SO(3)median angle error (deg): 28.7 -> 17.1Improved
Add synthetic training dataMatrix Fisher orientation estimation · Pascal3D+median angle error (deg): 8.90 -> 8.10Improved
Add data augmentation (ablation)Matrix Fisher orientation estimation · Pascal3D+median angle error (deg): 10.1 -> 8.9Improved
Replace cropping with homography warping (ablation)Matrix Fisher orientation estimation · Pascal3D+median angle error (deg): 10.0 -> 10.1No clear effect
Add a class embedding (ablation)Matrix Fisher orientation estimation · Pascal3D+median angle error (deg): 8.9 -> 8.9No clear effect
Add synthetic training dataMatrix Fisher orientation estimation · Pascal3D+ dining table classmedian angle error (deg): 8.4 -> 10.6Hurt

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