Research map / Detection and segmentation / CNN features
MixSignGraph: A Sign Sequence is Worth Mixed Graphs of Nodes
Read the paperCheck your idea against 108,967 papers
AI-extracted summary. Written by AI from the paper's abstract only. Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.
Proposes MixSignGraph, a graph-based sign language model that represents sign sequences as mixed graphs of nodes, building on the SignGraph approach as an alternative to CNN backbones.
Problem
CNN-based backbones pretrained on generic vision tasks (object detection, image recognition) capture contours and textures but may miss sign-related features needed for sign language understanding.
Approach
Represents a sign sequence as mixed graphs of nodes to capture sign-related features that CNN backbones transferred from generic vision tasks tend to miss; extends the SignGraph model.
Experimental setup
Baselines: SignGraph
Builds on: SignGraph
Closest papers by meaning
- SignGraph: A Sign Sequence is Worth Graphs of Nodes (CVPR 2024 · similarity 0.95)
- VSNet: Focusing on the Linguistic Characteristics of Sign Language (CVPR 2025 · similarity 0.87)
- Bridging Sign and Spoken Languages: Pseudo Gloss Generation for Sign Language Translation (NeurIPS 2025 · similarity 0.87)
- Lost in Translation, Found in Embeddings: Sign Language Translation and Alignment (NeurIPS 2026 · similarity 0.86)
- Improving Gloss-free Sign Language Translation by Reducing Representation Density (NeurIPS 2024 · similarity 0.86)
- TCNet: Continuous Sign Language Recognition from Trajectories and Correlated Regions (AAAI 2024 · similarity 0.86)
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.