AI
Isharah-Selfie: Continuous Sign Language Recognition Dataset for One-handed Signing
Publication: cs.AI updates on arXiv.org
Author: Ahmed Abul Hasanaath, Murtadha Aljubran, Sarah Alyami, Muhammad Haris Khan, Hamzah Luqman
Source published: 27/10/2026, 04:00:00 UTC
Briefing created: 22/11/2026, 16:51:05 UTC
Based on a feed excerpt; full article content has not been acquired. Source evidence retrieved 28/10/2026, 02:33:24 UTC.
Briefings
This briefing presents the Isharah-Selfie dataset, a large-scale Arabic Sign Language (ArSL) collection of 8,997 video clips from 16 signers using front-facing smartphones. The dataset features 1,493 unique sentences across healthcare, education, and transportation domains, capturing realistic one-handed, close-range mobile communication. It includes gloss sequences for continuous sign language recognition and Arabic sentence translations. Benchmarks show current methods struggle with selfie-style signing, especially under unseen-sentence evaluation, prompting the need for robust, user-centered technologies.
Key points
- The Isharah-Selfie dataset contains 8,997 video clips from 16 signers using front-facing smartphones.
- The dataset includes 1,493 unique sentences covering healthcare, education, and transportation domains.
Context
This dataset addresses a critical gap in Arabic Sign Language recognition benchmarks, which are typically collected in controlled studio settings. Unlike conventional datasets, Isharah-Selfie captures realistic mobile communication with one-handed signing, close-range viewpoints, and heterogeneous device resolutions, reflecting everyday usage scenarios.
Limitations
The evidence is limited to the arXiv preprint arXiv:2610.33200v1, which may not reflect final publication details. Evaluation protocols for signer-independent and unseen-sentences are defined but not yet validated against full-scale benchmarks. Results regarding method performance remain preliminary and subject to future verification.
Generated by Bittrees-hosted qwen35-2b sha256:aaf42c8b7c3c on 22/11/2026, 16:51:05 UTC. Based on publisher-provided feed evidence; may not cover the full original article. Check the source for details.
View archived briefing on IPFS
bafybeifzsrci3ll6ie7tcbx4vmieu34copc5bahdg7gkxxbmqileykywgy