AI Paper Atlas

Research map / Detection and segmentation / CNN features

Matching Is Not Enough: A Two-Stage Framework for Category-Agnostic Pose Estimation

Min Shi, Zihao Huang, Xianzheng Ma, Xiaowei Hu, Zhiguo Cao

CVPR 2023 · Detection and segmentation › CNN features · approach: Box regression and loss design · 22 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
replace one-stage matching with a two-stage proposal refinement decoder (ablation)CapeFormer · MP-100 split 1 test, 1-shotPCK: 85.32 -> 89.45Improved
add query-support joint refine encoder instead of DETR encoder (ablation)CapeFormer · MP-100 split 1 test, 1-shotPCK: 80.32 -> 82.86Improved
add fixed sinusoidal support keypoint identifier (ablation)CapeFormer · MP-100 split 1 test, 1-shotPCK: 82.86 -> 85.32Improved
use heatmap supervision together with expectation loss (ablation)CapeFormer similarity loss · MP-100 split 1 test, 1-shotPCK: 88.74 -> 89.45Improved
remove expectation loss and use heatmap supervision only (ablation)CapeFormer similarity loss · MP-100 split 1 test, 1-shotPCK: 89.45 -> 88.74Hurt
remove heatmap loss and use expectation supervision only (ablation)CapeFormer similarity loss · MP-100 split 1 test, 1-shotPCK: 89.41 -> 89.45No clear effect
replace fixed sinusoidal identifier with learnable embedding (ablation)CapeFormer · MP-100 split 1 test, 1-shotPCK: 89.45 -> 88.96Hurt
use CapeFormer instead of POMNetMP-100 · 1-shot, AveragePCK: 79.70 -> 85.31Improved
use CapeFormer instead of POMNetMP-100 · 5-shot, AveragePCK: 80.71 -> 89.30Improved
increase support examples from 1-shot to 5-shotCapeFormer · MP-100, AveragePCK: 85.31 -> 89.30Improved
use CapeFormer instead of POMNetMP-100 cross-super-category · 1-shot VehiclePCK: 34.92 -> 45.40Improved

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