Research map / Graph learning / Knowledge and temporal graphs
HyperSign: Saliency-Aware Spatial Graphs and Temporal Hypergraphs for Continuous Sign Language Recognition
Weiyi Ye, Xu-Hua Yang, Dong Wei, Gang-Feng Ma, Yujiao Huang, Xiao-Xin Li
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 changed | Where | Measured | Effect |
|---|---|---|---|
| add SASGM (ablation) | HyperSign · PHOENIX14 Dev | WER: 21.3 -> 17.5 | Improved |
| add THM (ablation) | HyperSign · PHOENIX14 Dev | WER: 21.3 -> 17.9 | Improved |
| add both SASGM and THM (ablation) | HyperSign · PHOENIX14 Test | WER: 21.1 -> 17.1 | Improved |
| saliency gating instead of K-NN graph (ablation) | SASGM · PHOENIX14 Test | WER: 17.5 -> 17.1 | Improved |
| add saliency to graph construction (vs standard K-NN) (ablation) | SASGM · PHOENIX14 Test | WER: 19.2 -> 17.1 | Improved |
| use fixed 4-frame window instead of 12 (ablation) | THM · PHOENIX14 Test | WER: 19.4 -> 17.1 | Improved |
| use 4-frame window instead of 2 (ablation) | THM · PHOENIX14 Test | WER: 17.7 -> 17.1 | Improved |
| combine outputs of all four stages (ablation) | HyperSign · PHOENIX14 Test | WER: 19.0 -> 17.1 | Improved |
| HyperSign vs single-cue SOTA OLMD | HyperSign · PHOENIX14-T Test | WER: 18.4 -> 18.2 | Improved |
| lambda 0.6 weighting of feature complexity vs motion (ablation) | SASGM · PHOENIX14 | WER | Improved |
Closest papers by meaning
- HyperSign: Hierarchical Hypergraph-based Co-occurrence Modeling for Sign Language Recognition and Translation (AAAI 2026 · similarity 0.93)
- TCNet: Continuous Sign Language Recognition from Trajectories and Correlated Regions (AAAI 2024 · similarity 0.89)
- Self-Emphasizing Network for Continuous Sign Language Recognition (AAAI 2023 · similarity 0.89)
- SignGraph: A Sign Sequence is Worth Graphs of Nodes (CVPR 2024 · similarity 0.89)
- Spatial-Temporal Multi-Cue Network for Continuous Sign Language Recognition (AAAI 2020 · similarity 0.89)
- Geo-Sign: Hyperbolic Contrastive Regularisation for Geometrically Aware Sign Language Translation (NeurIPS 2025 · similarity 0.87)
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.
