Existing sign language recognition and translation benchmarks are mostly collected in controlled settings with fixed or studio cameras, which limits their ability to reflect everyday mobile communication. This gap is especially evident for Arabic Sign Language (ArSL), the primary sign language across Arab countries.
We introduce the Isharah‑Selfie dataset, collected entirely with front‑facing smartphone cameras under unconstrained selfie conditions. The dataset comprises 8,997 video clips from 16 signers, covering 1,493 unique sentences across domains such as healthcare, education, and transportation.
Unlike conventional datasets dominated by fixed‑camera, two‑handed signing, Isharah‑Selfie captures realistic mobile communication characteristics: one‑handed signing, close‑range viewpoints, partial body visibility, camera motion, unconstrained environments, and heterogeneous device resolutions.
Each video is annotated with gloss sequences for continuous sign language recognition (CSLR) and Arabic sentence translations for sign language translation (SLT).
We define two evaluation protocols—signer‑independent and unseen‑sentence—to assess generalization to unseen signers and novel sentence compositions.
Extensive benchmarks using RGB‑based and pose‑based features show that current CSLR and SLT methods struggle under selfie‑style conditions, especially in unseen‑sentence evaluation where compositional generalization remains weak.
By releasing Isharah‑Selfie, we aim to support research toward robust, user‑centered sign language technologies that operate reliably in realistic smartphone‑based communication settings. The dataset is available at https://isharah-selfie.github.io/.
Review