no code implementations • 17 Apr 2024 • Harry Walsh, Ben Saunders, Richard Bowden
Sign languages, often categorised as low-resource languages, face significant challenges in achieving accurate translation due to the scarcity of parallel annotated datasets.
no code implementations • 8 Aug 2023 • Harry Walsh, Ozge Mercanoglu Sincan, Ben Saunders, Richard Bowden
As a result, research has turned to TV broadcast content as a source of large-scale training data, consisting of both the sign language interpreter and the associated audio subtitle.
no code implementations • SLTAT (LREC) 2022 • Harry Walsh, Ben Saunders, Richard Bowden
We use language models such as BERT and Word2Vec to create better sentence level embeddings, and apply several tokenization techniques, demonstrating how these improve performance on the low resource translation task of Text to Gloss.
1 code implementation • CVPR 2022 • Ben Saunders, Necati Cihan Camgoz, Richard Bowden
To learn sign co-articulation, we propose a novel Frame Selection Network (FS-Net) that improves the temporal alignment of interpolated dictionary signs to continuous signing sequences.
no code implementations • SLTAT (LREC) 2022 • Ben Saunders, Necati Cihan Camgoz, Richard Bowden
Recent approaches to Sign Language Production (SLP) have adopted spoken language Neural Machine Translation (NMT) architectures, applied without sign-specific modifications.
no code implementations • ICCV 2021 • Ben Saunders, Necati Cihan Camgoz, Richard Bowden
Using a progressive transformer for the translation sub-task, we propose a novel Mixture of Motion Primitives (MoMP) architecture for sign language animation.
no code implementations • 22 Jul 2021 • Ben Saunders, Necati Cihan Camgoz, Richard Bowden
To tackle SLVA, we propose AnonySign, a novel automatic approach for visual anonymisation of sign language data.
no code implementations • 5 May 2021 • Necati Cihan Camgoz, Ben Saunders, Guillaume Rochette, Marco Giovanelli, Giacomo Inches, Robin Nachtrab-Ribback, Richard Bowden
Computational sign language research lacks the large-scale datasets that enables the creation of useful reallife applications.
no code implementations • 11 Mar 2021 • Ben Saunders, Necati Cihan Camgoz, Richard Bowden
Sign languages are multi-channel visual languages, where signers use a continuous 3D space to communicate. Sign Language Production (SLP), the automatic translation from spoken to sign languages, must embody both the continuous articulation and full morphology of sign to be truly understandable by the Deaf community.
no code implementations • 19 Nov 2020 • Ben Saunders, Necati Cihan Camgoz, Richard Bowden
To be truly understandable and accepted by Deaf communities, an automatic Sign Language Production (SLP) system must generate a photo-realistic signer.
no code implementations • 27 Aug 2020 • Ben Saunders, Necati Cihan Camgoz, Richard Bowden
Sign Languages are rich multi-channel languages, requiring articulation of both manual (hands) and non-manual (face and body) features in a precise, intricate manner.
1 code implementation • ECCV 2020 • Ben Saunders, Necati Cihan Camgoz, Richard Bowden
The goal of automatic Sign Language Production (SLP) is to translate spoken language to a continuous stream of sign language video at a level comparable to a human translator.