AI Paper Atlas

Research map / Video / Video segmentation and tracking

CoSign: Exploring Co-occurrence Signals in Skeleton-based Continuous Sign Language Recognition

Peiqi Jiao, Yuecong Min, Yanan Li, Xiaotao Wang, Lei Lei, Xilin Chen

ICCV 2023 · Video › Video segmentation and tracking · approach: Sign language and physiological video analysis · 68 citations (OpenAlex)

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 changedWhereMeasuredEffect
add group-specific GCN modules (ablation)skeleton baseline · PHOENIX14 DevWER (%): 24.3 -> 22.8Improved
add group-specific centralization (ablation)group-specific GCN · PHOENIX14 DevWER (%): 22.8 -> 21.8Improved
add group dropout (ablation)CoSign-1s · PHOENIX14 DevWER (%): 21.8 -> 21.2Improved
add complementary regularization (ablation)CoSign-1s with group dropout · PHOENIX14 DevWER (%): 21.2 -> 21.0Improved
add complementary masks (ablation)CoSign-2s · PHOENIX14 DevWER (%): 20.2 -> 19.7Improved
gradual fusion instead of late fusion of skeleton and motion (ablation)CoSign-2s · PHOENIX14 DevWER (%): 21.0 -> 20.7Improved
add motion stream to skeleton stream (ablation)CoSign-1s · PHOENIX14 DevWER (%): 21.8 -> 20.7Improved
vary dropout mask clip length from 25 to 100 (ablation)CoSign-1s · PHOENIX14 DevWER (%): 21.2 -> 21.4No clear effect
use skeleton-based CoSign-2s instead of SMKD video modelPHOENIX14 training cost · PHOENIX14training time: ~19.5 h -> ~4.5 hImproved
use CoSign-2s on signer-independent splitPHOENIX14-SI · PHOENIX14-SI TestWER (%): 39.0 -> 29.6Improved
use CoSign-1s instead of baseline for inference speedinference · PHOENIX14inference speed (seq/s): 26.1 -> 18.8Hurt

Closest papers by meaning

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.

Open scime draft

Get the next scime Dispatch and a monthly field update

Free. We send a confirmation link first; every email has a one-click unsubscribe. Privacy