AI Paper Atlas

Research map / Label-efficient and robust learning / Recognition and long-tail

SoftHGNN: Soft Hypergraph Neural Networks for General Visual Recognition

Mengqi Lei, Yihong Wu, Siqi Li, Xinhu Zheng, Juan Wang, Shaoyi Du, Yue Gao

IJCV 2026 · Label-efficient and robust learning › Recognition and long-tail · approach: Decoupled and calibrated long-tail training · 11 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 SoftHGNN block at encoder endPVT-Tiny · ImageNet-100Top-1: 82.7 -> 86.1Improved
add SoftHGNN block at encoder endSwinT-Tiny · ImageNet-100Top-1: 88.0 -> 90.1Improved
add sparse hyperedge selection to SoftHGNNPVT-Tiny · ImageNet-100Top-1: 86.1 -> 86.8Improved
add sparse hyperedge selection to SoftHGNNSwinT-Tiny · ImageNet-100Top-1: 90.1 -> 90.9Improved
add k-NN HGNN blockPVT-Tiny · ImageNet-100Top-5: 95.4 -> 95.2Hurt
add self-attention blockPVT-Tiny · ImageNet-100Top-1: 82.7 -> 83.0No clear effect
add SoftHGNN block at encoder endPVT-Tiny · ImageNet-1kTop-1: 74.8 -> 75.6Improved
add sparse hyperedge selection to SoftHGNNPVT-Tiny · ImageNet-1kTop-1: 75.6 -> 75.9Improved
add SoftHGNN to multi-scale fusionCCTrans Twins-PCPVT · ShanghaiTech Part-AMAE: 54.8 -> 52.8Improved
add SoftHGNN to multi-scale fusionCCTrans PVT · ShanghaiTech Part-BMAE: 7.4 -> 6.8Improved
add SoftHGNN neck moduleYOLO11-N · MS COCOAP50:95: 38.2 -> 40.2Improved
add SoftHGNN neck moduleYOLOv12-M · MS COCOAP50:95: 51.9 -> 52.2Improved

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