From a8c3d75976ee9093005b2f7fb92ff7b0fd467662 Mon Sep 17 00:00:00 2001 From: Amit Moryossef Date: Tue, 28 Apr 2026 09:19:04 +0000 Subject: [PATCH 1/2] Add Malaia et al. 2024 on radar-based sign language corpora Cites the SignLang 2024 position paper proposing radar sensing as a non-intrusive, privacy-preserving alternative to motion capture and 2D video for capturing 3D articulator motion in sign language corpora. Co-Authored-By: Claude Opus 4.7 (1M context) --- src/index.md | 1 + src/references.bib | 20 ++++++++++++++++++++ 2 files changed, 21 insertions(+) diff --git a/src/index.md b/src/index.md index 21ead94..ae8a60f 100644 --- a/src/index.md +++ b/src/index.md @@ -189,6 +189,7 @@ This technique has been extensively used in the field of computer vision to esti where the goal is to determine the spatial configuration of the body at each point in time. Although high-quality pose estimation can be achieved using motion capture equipment, such methods are often expensive and intrusive. As a result, estimating pose from videos has become the preferred method in recent years [@pose:pishchulin2012articulated;@pose:chen2017adversarial;@pose:cao2018openpose;@pose:alp2018densepose]. +As a non-intrusive alternative, @malaia-etal-2024-capturing argue that radar sensing can capture 3D articulator motion at high spatial and temporal resolution, preserving depth information lost in 2D video while only recording kinematic parameters and thus inherently protecting signer identity. Compared to video representations, accurate skeletal poses have a lower complexity and provide a semi-anonymized representation of the human body, while observing relatively low information loss. However, they remain a continuous, multidimensional representation that is not adapted to most NLP models. diff --git a/src/references.bib b/src/references.bib index 226750a..ce148e3 100644 --- a/src/references.bib +++ b/src/references.bib @@ -4711,3 +4711,23 @@ @inproceedings{schulder-etal-2024-signs url = {https://aclanthology.org/2024.signlang-1.38}, year = {2024} } + +@inproceedings{malaia-etal-2024-capturing, + title = "Capturing Motion: Using Radar to Build Better Sign Language Corpora", + author = "Malaia, Evie and + Borneman, Joshua and + Gurbuz, Sevgi", + editor = "Efthimiou, Eleni and + Fotinea, Stavroula-Evita and + Hanke, Thomas and + Hochgesang, Julie A. and + Mesch, Johanna and + Schulder, Marc", + booktitle = "Proceedings of the LREC-COLING 2024 11th Workshop on the Representation and Processing of Sign Languages: Evaluation of Sign Language Resources", + month = may, + year = "2024", + address = "Torino, Italia", + publisher = "ELRA and ICCL", + url = "https://aclanthology.org/2024.signlang-1.23/", + pages = "213--218" +} From 2da65c7636a12c99dc9a32d8b7fb63800ecc7386 Mon Sep 17 00:00:00 2001 From: AmitMY Date: Tue, 28 Apr 2026 12:19:28 +0000 Subject: [PATCH 2/2] Add new ###### Radar Sensing representation subsection (review feedback) --- src/index.md | 5 ++++- 1 file changed, 4 insertions(+), 1 deletion(-) diff --git a/src/index.md b/src/index.md index ae8a60f..2289bdc 100644 --- a/src/index.md +++ b/src/index.md @@ -189,10 +189,13 @@ This technique has been extensively used in the field of computer vision to esti where the goal is to determine the spatial configuration of the body at each point in time. Although high-quality pose estimation can be achieved using motion capture equipment, such methods are often expensive and intrusive. As a result, estimating pose from videos has become the preferred method in recent years [@pose:pishchulin2012articulated;@pose:chen2017adversarial;@pose:cao2018openpose;@pose:alp2018densepose]. -As a non-intrusive alternative, @malaia-etal-2024-capturing argue that radar sensing can capture 3D articulator motion at high spatial and temporal resolution, preserving depth information lost in 2D video while only recording kinematic parameters and thus inherently protecting signer identity. Compared to video representations, accurate skeletal poses have a lower complexity and provide a semi-anonymized representation of the human body, while observing relatively low information loss. However, they remain a continuous, multidimensional representation that is not adapted to most NLP models. +###### Radar Sensing {-} +captures 3D articulator motion through electromagnetic returns rather than visual frames, recording only kinematic parameters and thus inherently protecting signer identity. +@malaia-etal-2024-capturing argue that radar can capture sign language motion at high spatial and temporal resolution, preserving depth information lost in 2D video as a non-intrusive alternative to motion capture. + ###### Written notation systems {-} represent signs as discrete visual features. Some systems are written linearly, and others use graphemes in two dimensions. While various universal [@writing:sutton1990lessons;@writing:prillwitz1990hamburg] and language-specific notation systems [@writing:stokoe1960sign;@writing:kakumasu1968urubu;@writing:bergman1977tecknad] have been proposed,