Research map / Video / Video segmentation and tracking
Signing at Scale: Learning to Co-Articulate Signs for Large-Scale Photo-Realistic Sign Language Production
Ben Saunders, Necati Cihan Camgoz, Richard Bowden
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 |
|---|---|---|---|
| Replace direct pose regression with learned temporal frame selection and alignment | Gloss-to-Pose production · PHOENIX14T back translation test | BLEU-4: 13.30 -> 21.10 | Improved |
| Use learned co-articulation rather than an interpolated dictionary sequence | Gloss-to-Pose production · PHOENIX14T back translation test | BLEU-4: 16.27 -> 18.78 | Improved |
| Add a hand keypoint loss (ablation) | SignGAN · C4A | Hand SSIM: 0.582 -> 0.605 | Improved |
| Add a hand keypoint loss (ablation) | SignGAN · C4A | FID: 39.33 -> 27.75 | Improved |
| Add a hand discriminator instead of a hand keypoint loss (ablation) | SignGAN · C4A | Hand SSIM: 0.582 -> 0.565 | Hurt |
| Add a hand discriminator instead of a hand keypoint loss (ablation) | SignGAN · C4A | SSIM: 0.743 -> 0.738 | Hurt |
| Generate photo-realistic signing videos instead of skeleton sequences | SignGAN · Deaf user comprehension evaluation | understandability rating: 3.2 -> 3.9 | Improved |
| Use FS-Net co-articulated sequences rather than the baseline SLP method | Sign language production · mDGS and PHOENIX14T comprehension preference | participant preference: 0% -> 95% | Improved |
| Use FS-Net co-articulation rather than isolated non-trimmed dictionary signs | Dictionary-sign production · Non-trimmed dictionary signs | smooth-transition preference: 0% -> 100% | Improved |
Closest papers by meaning
- Scaling up Multimodal Pre-Training for Sign Language Understanding (TPAMI 2025 · similarity 0.91)
- Advanced Sign Language Video Generation with Compressed and Quantized Multi-Condition Tokenization (NeurIPS 2025 · similarity 0.89)
- Lost in Translation, Found in Embeddings: Sign Language Translation and Alignment (NeurIPS 2026 · similarity 0.89)
- Neural Sign Actors: A Diffusion Model for 3D Sign Language Production from Text (CVPR 2024 · similarity 0.89)
- Sign-IDD: Iconicity Disentangled Diffusion for Sign Language Production (AAAI 2025 · similarity 0.89)
- VSNet: Focusing on the Linguistic Characteristics of Sign Language (CVPR 2025 · similarity 0.88)
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.
