AI Paper Atlas

Research map / Detection and segmentation / CNN features

Attentional Pyramid Pooling of Salient Visual Residuals for Place Recognition

Guohao Peng, Jun Zhang, Heshan Li, Danwei Wang

ICCV 2021 · Detection and segmentation › CNN features · approach: Local features and context in CNNs · 60 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 local weighting scheme to NetVLAD (ablation)NetVLAD · Tokyo24/7 (4096-D)Recall@1 (%): 68.9 -> 70.8Improved
add semantic constrained initialization (ablation)APPSVR (LW) · Tokyo24/7 (4096-D)Recall@1 (%): 70.8 -> 72.1Improved
add parametric normalization (ablation)APPSVR (LW) · Tokyo24/7 (4096-D)Recall@1 (%): 70.8 -> 73.0Improved
add parametric normalization (ablation)APPSVR-SC · Tokyo24/7 (4096-D)Recall@1 (%): 72.1 -> 73.7Improved
add parametric normalization (ablation)APPSVR-SC · Pitts250k-test (4096-D)Recall@1 (%): 87.8 -> 88.1Improved
add 2-scale attentional pyramid pooling (ablation)APPSVR-SC-PN · Tokyo24/7 (4096-D)Recall@1 (%): 73.7 -> 75.2Improved
add 2-scale attentional pyramid pooling (ablation)APPSVR-SC-PN · Pitts30k-test (4096-D)Recall@1 (%): 86.4 -> 87.3Improved
increase pyramid scales from 2 to 3 (ablation)APPSVR-SC-PN-L2 · Tokyo24/7 (4096-D)Recall@1 (%): 75.2 -> 77.1Improved
increase pyramid scales from 2 to 3 (ablation)APPSVR-SC-PN-L2 · Pitts250k-test (4096-D)Recall@1 (%): 88.5 -> 88.8Improved
add semantic constrained initialization (ablation)APPSVR-SC-PN (Tokyo Recall@5, no AP) · Tokyo24/7 (4096-D)Recall@5 (%): 80.0 -> 83.2Improved
add parametric normalization to other VLAD variants (ablation)NetVLAD, GhostVLAD, CRN, etc. · Pittsburgh and TokyoRecall: stable improvements for all baseline modelsImproved
use full APPSVR instead of NetVLADNetVLAD · Pitts250k-test (512-D)Recall@1 (%): 83.3 -> 85.3Improved

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