Research map / Large language models / Language, safety and interpretability
C2SLR: Consistency-Enhanced Continuous Sign Language Recognition
Ronglai Zuo, Brian Mak
Check 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 keypoint-guided spatial attention consistency (SAC) (ablation) | VLT · PHOENIX-2014 | WER: 21.5 -> 20.8 | Improved |
| Add sentence embedding consistency (SEC) (ablation) | VLT · PHOENIX-2014 | WER: 21.5 -> 20.9 | Improved |
| Add SAC and SEC together (ablation) | VLT · PHOENIX-2014 | WER: 21.5 -> 20.4 | Improved |
| Add SAC and SEC together (ablation) | VTB · PHOENIX-2014 | WER: 25.0 -> 22.6 | Improved |
| Add SAC and SEC together (ablation) | CT · PHOENIX-2014 | WER: 26.1 -> 24.5 | Improved |
| Add spatial attention without its heatmap-guidance loss (ablation) | VLT · PHOENIX-2014 | WER: 21.5 -> 21.4 | No clear effect |
| Add dynamic channel weights to the spatial attention module (ablation) | VLT + SAC · PHOENIX-2014 | WER: 21.3 -> 20.8 | Improved |
| Add heatmap post-processing and refinement (ablation) | VLT + SAC · PHOENIX-2014 | WER: 21.7 -> 20.8 | Improved |
| Use a Transformer plus depth-wise TCN sentence embedding extractor rather than Transformer alone (ablation) | VLT + SEC · PHOENIX-2014 | WER: 21.1 -> 20.9 | Improved |
| Use a Transformer-based sentence embedding extractor rather than BiLSTM (ablation) | VLT + SEC · PHOENIX-2014 | WER: 21.3 -> 20.9 | Improved |
| Add negative sampling to SEC (ablation) | VLT + SEC · PHOENIX-2014 | WER: 21.5 -> 20.9 | Improved |
| Enforce consistency at the sentence level instead of the frame level (ablation) | VLT · PHOENIX-2014 | WER: 21.6 -> 20.9 | Improved |
Later papers that compared with this method
2 later accepted papers report C2SLR as a baseline in their results tables; 2 of them report a better number on most of the results they share with it.
- Cross-Sentence Gloss Consistency for Continuous Sign Language Recognition (AAAI 2024 · better on 1 of 1 shared result)
- Improving Continuous Sign Language Recognition with Cross-Lingual Signs (ICCV 2023 · better on 1 of 1 shared result)
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
- Iterative Reference Driven Metric Learning for Signer Independent Isolated Sign Language Recognition (ECCV 2016 · similarity 0.86)
- LexSign: Learning Sign Language from Lexical Descriptions (IJCV 2026 · similarity 0.86)
- Cross-Sentence Gloss Consistency for Continuous Sign Language Recognition (AAAI 2024 · similarity 0.85)
- TCNet: Continuous Sign Language Recognition from Trajectories and Correlated Regions (AAAI 2024 · similarity 0.85)
- OLMD: Orientation-aware Long-term Motion Decoupling for Continuous Sign Language Recognition (AAAI 2025 · similarity 0.84)
- A Wireframe-Based Approach for Classifying and Acquiring Proficiency in the American Sign Language (Student Abstract) (AAAI 2024 · similarity 0.82)
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.
