AI Paper Atlas

Research map / Architectures and efficiency / Attention and transformers

ASpanFormer: Detector-Free Image Matching with Adaptive Span Transformer

Hongkai Chen, Zixin Luo, Lei Zhou, Yurun Tian, Mingmin Zhen, Tian Fang, David McKinnon, Yanghai Tsin, Long Quan

ECCV 2022 · Architectures and efficiency › Attention and transformers · approach: Efficient image transformers · 302 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 hierarchical global-local attention with fixed 13 px span over global-only attention (ablation)ASpanFormer · ScanNetpose AUC@5deg: 22.65 -> 24.85Improved
make local attention span adaptive from flow uncertainty instead of fixed 13 px (ablation)ASpanFormer · ScanNetpose AUC@5deg: 24.85 -> 25.61Improved
make local attention span adaptive (AUC@20) (ablation)ASpanFormer · ScanNetpose AUC@20deg: 62.71 -> 63.33Improved
add learnable softmax temperatures per level (ablation)ASpanFormer · ScanNetpose AUC@20deg: 62.3 -> 63.3Improved
use 3x3 conv FFN in place of self attention plus MLP FFN (ablation)ASpanFormer · ScanNetpose AUC@5deg: 24.8 -> 25.6Improved
add normalized positional encoding when test resolution differs from training (ablation)ASpanFormer · MegaDepthpose AUC@5deg: 52.8 -> 55.3Improved
add normalized positional encoding (flow accuracy) (ablation)ASpanFormer · MegaDepthflow accuracy: 22.6 -> 72.3Improved
increase GLA iterations for flow regression (ablation)ASpanFormer · ScanNetflow accuracy <6px (%): 69.1 -> 72.3Improved
increase GLA iterations: uncertainty of unmatchable pixels (ablation)ASpanFormer · ScanNetmean 5 sigma (px) unmatchable: 19.4 -> 27.1Improved
replace linear attention (LoFTR) with ASpanFormer attention module: runtimeASpanFormer vs LoFTR · ScanNet 640x480total runtime (ms): 97.7 -> 113.5Hurt
replace LoFTR with ASpanFormer in HLocHLoc · InLoc DUC2localization accuracy (1m,10deg): 85.5 -> 81.7Hurt
replace SuperGlue with ASpanFormer in HLocHLoc · InLoc DUC1localization accuracy (0.25m,2deg): 49.0 -> 51.5Improved

Later papers that compared with this method

8 later accepted papers report ASpanFormer as a baseline in their results tables; 5 of them report a better number on most of the results they share with it.

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