AI Paper Atlas

Research map / Large language models / Language, safety and interpretability

CiCo: Domain-Aware Sign Language Retrieval via Cross-Lingual Contrastive Learning

Yiting Cheng, Fangyun Wei, Jianmin Bao, Dong Chen, Wenqiang Zhang

CVPR 2023 · Large language models › Language, safety and interpretability · approach: Sign and cross-lingual translation · 33 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
Use CiCo instead of SPOT-ALIGN's strongest SA-COMB baselineSign language retrieval · How2Sign testT2V R@1: +22.4Improved
Use CiCo instead of SPOT-ALIGN's strongest SA-COMB baselineSign language retrieval · PHOENIX-2014T testV2T R@1: +17.1Improved
Add the domain-aware/domain-agnostic sign encoder (ablation)CiCo baseline · How2Sign testT2V R@1: +1.8Improved
Add cross-lingual contrastive learning with fine-grained sign-to-word alignment (ablation)CiCo with sign encoder · How2Sign testT2V R@1: +20.7Improved
Add random-swap text augmentation (ablation)CiCo without text augmentation · How2Sign testT2V R@1: +2.6Improved
Apply random-delete text augmentation (ablation)CiCo without text augmentation · How2Sign testT2V R@1: 54.0 -> 52.7Hurt
Apply synonym-replacement text augmentation (ablation)CiCo without text augmentation · How2Sign testV2T R@1: 50.0 -> 47.5Hurt
Initialize vision and text Transformers from CLIP rather than randomly (ablation)CiCo · How2Sign testT2V R@1: +11.4Improved
Freeze shallow text Transformer layers instead of fine-tuning all layers (ablation)CiCo · How2Sign testT2V R@1: 56.6 -> 54.4Hurt
Use softmax mapping instead of mean or max for fine-grained sign-to-word alignment (ablation)CiCo CLCL · How2Sign testT2V R@1: 33.1 / 42.2 -> 56.6Improved
Combine domain-agnostic and domain-aware features with weighted sum rather than average (ablation)CiCo sign encoder · How2Sign testT2V R@1: 54.7 -> 56.6Improved
Use a longer maximum sign-clip feature sequence (ablation)CiCo · How2Sign testT2V R@1: 17.3 -> 38.2 -> 50.9 -> 53.6 -> 56.6Improved

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