EAPSI: Understanding attention during the early stages of language acquisition using a new tracking eye gaze method

当涉及到了解儿童如何学习他们的第一个单词时，理解视觉注意力是如何被引导的尤为重要。当孩子对世界的看法不同时，学习的对象在很多方面都是不同的。当接触到人类语言时，孩子们可以选择凝视各种各样的物品，其中一些物品可能会为他们所看到的物理事物（视觉信息）提供基础词汇（听觉信息）。这个项目将应用最先进的技术，用于研究儿童在学习时看的地方。本研究旨在为儿童眼动追踪研究中遇到的问题提供解决方案。日本京都大学的西田博士拥有知识和专业知识，通过实施一种有前途的注视点估计技术来帮助实现这一目标。目前观察学习过程中注意力如何集中的方法通常包括在实验条件下使用眼球注视追踪。目前，有多种硬件选项可用于完成此任务，但各有其独特的优点和缺点。移动、可穿戴式眼动追踪解决方案在限制较少的环境中提供了更大的灵活性，但需要更长的设置时间、调整和校准。这增加了实验过程的整体侵入性，并经常导致孩子意识到这些装置。眼球追踪硬件没有连接到孩子身上（例如，一个专门的视频监视器），通常有更精确的凝视方向测量，但缺乏在三维环境中测量凝视的能力。该项目将使用的新方法是利用计算机视觉算法在超分辨率眼睛图像上分析角膜反射，以更好地适应儿童行为任务。它有可能大大减少在深度变化条件下获得准确和精确数据的复杂性，从而理解语言学习过程早期阶段注意力的作用。NSF EAPSI奖是与日本科学促进会合作资助的。

When it comes to knowing how children learn their first words, understanding how visual attention is guided is especially important. Objects being learned can vary in many ways when taking a child's view of the world. When exposed to human speech, children have the option to gaze upon a wide range of items, some of which are potentially informative for grounding words (auditory information) to physical things they see (visual information). This project will apply cutting edge techniques used for study where children look when they learn. This research aims to develop a solution for the problems we encounter in child-directed eye tracking research. Dr. Nishida at Kyoto University, Japan, has the knowledge and expertise to help with achieving this goal by implementing a promising technique for point of gaze estimation.Current methods for observing how attention is focused during learning typically involve the use of eye gaze tracking under experimental conditions. Currently, multiple hardware options exist for accomplishing this task, but have their own unique set of advantages and disadvantages. Mobile, wearable eye tracking solutions provide more flexibility in less constrained settings but require longer setup times, adjustments, and calibration. This increases the overall intrusiveness of the experimental procedure and often leads to the child being aware of such devices. Eye-tracking hardware not attached to the child (e.g., a specialized video monitor) typically have more precise measurements of gaze orientation but lack the ability to measure gaze in three-dimensional environments. The new method that will be used in this project, analyzes corneal reflections using computer vision algorithms on super resolution eye images to better accommodate child behavioral tasks. It has the potential to significantly reduce the complexities involved in obtaining accurate and precise data under depth-varying conditions for understanding the role of attention during the early stages of the language learning process. This NSF EAPSI award is funded in collaboration with the Japan Society for the Promotion of Science.