Research map / Video / Video segmentation and tracking
Spatial-Temporal Multi-Cue Network for Continuous Sign Language Recognition
Read the paperCheck your idea against 108,967 papers
AI-extracted summary. Written by AI from the paper's title only (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 multi-cue spatial features with SMC (ablation) | VGG+1D-CNN · PHOENIX-2014 Test | WER: 25.5 -> 22.4 | Helped |
| replace 1D-CNN temporal modelling with TMC without joint loss optimization (ablation) | SMC+1D-CNN · PHOENIX-2014 Test | WER: 22.4 -> 21.9 | Helped |
| add joint loss optimization to TMC (ablation) | SMC+TMC w/o JL · PHOENIX-2014 Test | WER: 21.9 -> 20.7 | Helped |
| add joint loss optimization to TMC (ablation) | SMC+TMC w/o JL · PHOENIX-2014 Dev | WER: 22.1 -> 21.1 | Helped |
| use the inter-cue path instead of a single intra-cue path (ablation) | TMC · PHOENIX-2014 Dev | WER: 25.0 -> 21.1 | Helped |
| use only the hand intra-cue path (ablation) | TMC · PHOENIX-2014 Dev | WER: 21.1 -> 30.5 | Hurt |
| use only the face intra-cue path (ablation) | TMC · PHOENIX-2014 Dev | WER: 21.1 -> 35.0 | Hurt |
| use only the pose intra-cue path (ablation) | TMC · PHOENIX-2014 Dev | WER: 21.1 -> 51.4 | Hurt |
| use a self-contained pose estimation branch rather than external pose estimation (ablation) | TMC pose estimation · PHOENIX-2014 Dev | WER: 21.7 -> 21.1 | Helped |
| use a self-contained pose estimation branch rather than an external VGG-11 pose model (ablation) | TMC pose estimation · PHOENIX-2014 inference | Time: 628ms -> 352ms | Helped |
| increase the auxiliary-loss weight slightly, excluding α=0 (ablation) | STMC · PHOENIX-2014 | WER | No clear effect |
| add STMC multi-cue learning | best reported prior method · CSL Split I | WER: 10.2 -> 2.1 | Helped |
Closest papers by meaning
- Self-Emphasizing Network for Continuous Sign Language Recognition (AAAI 2023 · similarity 0.90)
- Recurrent Convolutional Neural Networks for Continuous Sign Language Recognition by Staged Optimization (CVPR 2017 · similarity 0.90)
- Distilling Cross-Temporal Contexts for Continuous Sign Language Recognition (CVPR 2023 · similarity 0.89)
- TCNet: Continuous Sign Language Recognition from Trajectories and Correlated Regions (AAAI 2024 · similarity 0.89)
- SignGraph: A Sign Sequence is Worth Graphs of Nodes (CVPR 2024 · similarity 0.88)
- OLMD: Orientation-aware Long-term Motion Decoupling for Continuous Sign Language Recognition (AAAI 2025 · similarity 0.85)
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.