A computer-vision approach for testing developmental changes in object manipulation

Krishna, A. and Kaplan, B. and Pope, J. and Todorovic, S. and Adolph, K. and Ossmy, Ori (2024) A computer-vision approach for testing developmental changes in object manipulation. In: UNSPECIFIED (ed.) 2024 IEEE International Conference on Development and Learning (ICDL). IEEE. ISBN 9798350348569.

Abstract

Object manipulation is a foundational behavior that emerges in infancy and improves with age. Research on children’s interactions with objects is a cornerstone for understanding cognitive and motor development. Traditionally, developmental researchers rely on human video annotation to analyze object interactions by categorizing behavioral events (e.g., “banging,” “constructing”). However, human video annotation cannot provide precise, quantitative details from moment to moment about the location and orientation of objects and the movements of each hand and finger during reach, grasp, and manipulation. To overcome the challengesin acquiring realtime, continuous, quantitative data, researchers turned to highspeed motion tracking and inertial measurement units—which require children to wear markers—and to instrumenting the objects. However, “wearables” present a new set of complications as they disrupt the natural spontaneity of children's movements and sensors may fail to accurately track changes in object position and orientation. Critically, only video data captures the subtleties and complexities of manual behavior and the surrounding context. Consequently, we devised a novel, video-based, marker- and sensor-free approach that enables real-time quantification of children’s coordination patterns during object interaction. We demonstrate the power of our approach in a tower-building study with children (2 to 8 years) and adults. This approach marks a paradigm shift in testing the evolving dynamics of object manipulation over development.

Metadata

Statistics

DownloadsShow export options

Export as JSONXMLCSV

Activity Overview

6 month trend

1Download

6 month trend

71Hits

Additional statistics are available via IRStats2.

Archive Staff Only (login required)

Edit/View Item