AI Paper Atlas

Research map / Detection and segmentation / CNN features

SAPA: Similarity-Aware Point Affiliation for Feature Upsampling

Hao Lu, Wenze Liu, Zixuan Ye, Hongtao Fu, Yuliang Liu, Zhiguo Cao

NeurIPS 2022 · Detection and segmentation › CNN features · approach: Local features and context in CNNs · 25 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 the upsampling stages with SAPA-GSegFormer-B1 · ADE20KmIoU: 42.82 -> 44.39Improved
Replace the upsampling stages with SAPA-GMaskFormer-Swin-Base · ADE20KmIoU: 53.53 -> 53.78Improved
Replace the upsampling stages with SAPA-GMask2Former-Swin-Base · ADE20KmIoU: 53.94 -> 55.22Improved
Use SAPA-B instead of CARAFEFaster R-CNN R50 · MS COCOAP: 38.6 -> 37.8Hurt
Replace the upsampling stages with SAPA-GBTS R50 · NYU Depth V2RMS: 0.419 -> 0.408Improved
Replace the upsampling stages with SAPA-GA2U Matting R34 · Adobe Composition-1k testSAD: 32.05 -> 30.98Improved
Add gated bilinear similarity rather than plain addition (ablation)SAPA on SegFormer-B1 · ADE20KmIoU: 43.2 -> 44.4Improved
Use exp normalization rather than no normalization (ablation)SAPA on SegFormer-B1 · ADE20KmIoU: 41.5 -> 43.2Improved
Use ReLU rather than exp normalization (ablation)SAPA on SegFormer-B1 · ADE20KmIoU: 43.2 -> 40.8Hurt
Use sigmoid rather than exp normalization (ablation)SAPA on SegFormer-B1 · ADE20KmIoU: 43.2 -> 42.8Hurt
Use softplus rather than exp normalization (ablation)SAPA on SegFormer-B1 · ADE20KmIoU: 43.2 -> 42.7Hurt
Increase kernel size from 5 to 7 (ablation)SAPA on SegFormer-B1 · ADE20KmIoU: 43.2 -> 42.5Hurt

Earlier methods this paper compares with

Named baselines from earlier accepted papers in this paper's results tables: better than 1 of 1 on most shared results.

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