AI Paper Atlas

Research map / Vision-language and multimodal / Human and embodied modeling

Dynamic Position-aware Network for Fine-grained Image Recognition

Shijie Wang, Haojie Li, Zhihui Wang, Wanli Ouyang

AAAI 2021 · Vision-language and multimodal › Human and embodied modeling · approach: Scene graph generation · 30 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 discriminative patch sampling to the BL baseline (ablation)BL on CUB-200-2011 · CUB-200-2011accuracy: +2.6%Improved
Add PEM and PRM to the sampled baseline (ablation)BL + Sample on CUB-200-2011 · CUB-200-2011accuracy: 87.1% -> 88.4%Improved
Add PAM to the PEM + PRM variant (ablation)BL + Sample + PEM + PRM on CUB-200-2011 · CUB-200-2011accuracy: 88.4% -> 89.3%Improved
Use graph depth 2 instead of depth 1 (ablation)DP-Net on CUB-200-2011 · CUB-200-2011accuracy: 88.2% -> 88.7%Improved
Use graph depth 3 instead of depth 2 (ablation)DP-Net on CUB-200-2011 · CUB-200-2011accuracy: 88.7% -> 89.3%Improved
Increase graph depth from 3 to 4 (ablation)DP-Net on CUB-200-2011 · CUB-200-2011accuracy: 89.3% -> 89.1%Hurt
Increase graph depth from 3 to 5 (ablation)DP-Net on CUB-200-2011 · CUB-200-2011accuracy: 89.3% -> 88.6%Hurt
Add DP-Net position-aware learning to the original feature branch (ablation)Original Feature Branch on CUB-200-2011 · CUB-200-2011accuracy: 87.1% -> 88.2%Improved
Use the position-aware feature branch instead of the original feature branch (ablation)DP-Net on CUB-200-2011 · CUB-200-2011accuracy: 88.2% -> 88.5%Improved
Combine the original and position-aware feature branches (ablation)DP-Net on CUB-200-2011 · CUB-200-2011accuracy: 88.2% -> 89.3%Improved
Use DP-Net instead of the strongest listed baselineFine-grained recognition · CUB-200-2011accuracy: 88.5% -> 89.3%Improved
Use DP-Net instead of the strongest listed baselineFine-grained recognition · Stanford Cars-196accuracy: 94.7% -> 94.8%Improved

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