<?xml version="1.0" encoding="utf-8" standalone="yes"?><rss version="2.0" xmlns:atom="http://www.w3.org/2005/Atom"><channel><title>Joint Attention | Arek Bialek</title><link>https://arekbialek.github.io/tag/joint-attention/</link><atom:link href="https://arekbialek.github.io/tag/joint-attention/index.xml" rel="self" type="application/rss+xml"/><description>Joint Attention</description><generator>Hugo Blox Builder (https://hugoblox.com)</generator><language>en-us</language><lastBuildDate>Wed, 29 Apr 2026 00:00:00 +0000</lastBuildDate><image><url>https://arekbialek.github.io/media/icon_hu0b7a4cb9992c9ac0e91bd28ffd38dd00_9727_512x512_fill_lanczos_center_3.png</url><title>Joint Attention</title><link>https://arekbialek.github.io/tag/joint-attention/</link></image><item><title>Automated Detection of Mutual Gaze and Joint Attention in Dual-Camera Settings via Dual-Stream Transformers</title><link>https://arekbialek.github.io/publication/arxiv-2026-mutual-gaze-joint-attention/</link><pubDate>Wed, 29 Apr 2026 00:00:00 +0000</pubDate><guid>https://arekbialek.github.io/publication/arxiv-2026-mutual-gaze-joint-attention/</guid><description>&lt;h2 id="automated-measurement-of-mutual-gaze-and-joint-attention">Automated measurement of mutual gaze and joint attention&lt;/h2>
&lt;p>How can researchers detect mutual gaze and joint attention when an infant and caregiver are recorded by different cameras? This preprint proposes a &lt;strong>dual-stream Transformer&lt;/strong> that processes two synchronized video views and combines gaze-related information to identify coordinated looking behavior.&lt;/p>
&lt;p>The model was evaluated using recordings of infant–caregiver interactions and, in the authors&amp;rsquo; comparisons, outperformed a convolutional baseline and a multimodal language model. The work offers a promising route toward reducing time-consuming manual coding while preserving the advantages of studying interaction in more naturalistic settings.&lt;/p>
&lt;p>&lt;em>This work is a preprint, not a peer-reviewed journal article.&lt;/em>&lt;/p>
&lt;p>&lt;a href="https://doi.org/10.48550/arXiv.2604.27105" target="_blank" rel="noopener">Read the preprint&lt;/a> · &lt;a href="https://arxiv.org/pdf/2604.27105" target="_blank" rel="noopener">Download the PDF&lt;/a>&lt;/p></description></item></channel></rss>