STARK: Spatio-Temporal Attention for Representation of Keypoints for Continuous Sign Language Recognition

Fuente: arXiv
Saved in:
Bibliographic Details
Main Authors: Patra, Suvajit, Samanta, Soumitra
Format: Preprint
Published: 2026
Subjects:
Online Access:
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1866912970746888192
author Patra, Suvajit
Samanta, Soumitra
author_facet Patra, Suvajit
Samanta, Soumitra
contents Continuous Sign Language Recognition (CSLR) is a crucial task for understanding the languages of deaf communities. Contemporary keypoint-based approaches typically rely on spatio-temporal encoding, where spatial interactions among keypoints are modeled using Graph Convolutional Networks or attention mechanisms, while temporal dynamics are captured using 1D convolutional networks. However, such designs often introduce a large number of parameters in both the encoder and the decoder. This paper introduces a unified spatio-temporal attention network that computes attention scores both spatially (across keypoints) and temporally (within local windows), and aggregates features to produce a local context-aware spatio-temporal representation. The proposed encoder contains approximately $70-80\%$ fewer parameters than existing state-of-the-art models while achieving comparable performance to keypoint-based methods on the Phoenix-14T dataset.
format Preprint
id arxiv_https___arxiv_org_abs_2603_16163
institution arXiv
publishDate 2026
record_format arxiv
spellingShingle STARK: Spatio-Temporal Attention for Representation of Keypoints for Continuous Sign Language Recognition
Patra, Suvajit
Samanta, Soumitra
Computer Vision and Pattern Recognition
Computation and Language
Continuous Sign Language Recognition (CSLR) is a crucial task for understanding the languages of deaf communities. Contemporary keypoint-based approaches typically rely on spatio-temporal encoding, where spatial interactions among keypoints are modeled using Graph Convolutional Networks or attention mechanisms, while temporal dynamics are captured using 1D convolutional networks. However, such designs often introduce a large number of parameters in both the encoder and the decoder. This paper introduces a unified spatio-temporal attention network that computes attention scores both spatially (across keypoints) and temporally (within local windows), and aggregates features to produce a local context-aware spatio-temporal representation. The proposed encoder contains approximately $70-80\%$ fewer parameters than existing state-of-the-art models while achieving comparable performance to keypoint-based methods on the Phoenix-14T dataset.
title STARK: Spatio-Temporal Attention for Representation of Keypoints for Continuous Sign Language Recognition
topic Computer Vision and Pattern Recognition
Computation and Language
url https://arxiv.org/abs/2603.16163