Research map / Video / Action recognition
Temporal Lift Pooling for Continuous Sign Language Recognition
Lianyu Hu, Liqing Gao, Zekang Liu, Wei Feng
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 |
|---|---|---|---|
| replace max pooling with TLP at both pooling locations | CSLR ResNet18 baseline · PHOENIX14 Dev | WER (%): 21.2 -> 19.7 | Improved |
| replace max pooling with TLP | CSLR ResNet18 baseline · PHOENIX14-T Test | WER (%): 22.8 -> 21.2 | Improved |
| replace max pooling with stochastic pooling | CSLR ResNet18 baseline · PHOENIX14 Dev | WER (%): 21.2 -> 22.2 | Hurt |
| replace max pooling with average pooling | CSLR ResNet18 baseline · PHOENIX14 Dev | WER (%): 21.2 -> 21.1 | No clear effect |
| add TLP to SqueezeNet backbone | SqueezeNet · PHOENIX14 Dev | WER (%): 23.2 -> 22.2 | Improved |
| add TLP to RegNetY-800Mf backbone | RegNetY-800Mf · PHOENIX14 Dev | WER (%): 21.3 -> 19.7 | Improved |
| add TLP to FCN | FCN · PHOENIX14 Dev | WER (%): 24.2 -> 22.4 | Improved |
| add TLP to baseline on CSL | CSLR baseline · CSL | WER (%): 7.3 -> 1.8 | Improved |
| add TLP at only first location (ablation) | CSLR baseline · PHOENIX14 Dev | WER (%): 21.2 -> 20.1 | Improved |
| add TLP at both locations rather than only second (ablation) | TLP · PHOENIX14 Dev | WER (%): 20.3 -> 19.7 | Improved |
| fuse high-pass d* with s* by sum instead of keeping only s* (ablation) | TLP · PHOENIX14 Dev | WER (%): 20.2 -> 19.7 | Improved |
| use residual component weighting instead of direct weighting (ablation) | TLP · PHOENIX14 Dev | WER (%): 20.9 -> 19.7 | Improved |
Closest papers by meaning
- TCNet: Continuous Sign Language Recognition from Trajectories and Correlated Regions (AAAI 2024 · similarity 0.89)
- Recurrent Convolutional Neural Networks for Continuous Sign Language Recognition by Staged Optimization (CVPR 2017 · similarity 0.88)
- Spatial-Temporal Multi-Cue Network for Continuous Sign Language Recognition (AAAI 2020 · similarity 0.88)
- Distilling Cross-Temporal Contexts for Continuous Sign Language Recognition (CVPR 2023 · similarity 0.87)
- OLMD: Orientation-aware Long-term Motion Decoupling for Continuous Sign Language Recognition (AAAI 2025 · similarity 0.86)
- KD-MSLRT: Lightweight Sign Language Recognition Model Based on Mediapipe and 3D to 1D Knowledge Distillation (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.
