Computer Vision in Human Developmental Research
Part of my work is dedicated to increasing scientific transparency. I joined the Databrary project – a free, online repository that supports video data sharing among behavioral scientists. Databrary breaks down economic and accessibility barriers for researchers by creating an online platform that allows for an unprecedented level of open data sharing and video reuse. Databrary also brings videos of children from underrepresented groups and regions to the larger science community.
We are currently developing AutoViDev—an automatic video-analysis tool that uses machine learning and computer vision to support video-based developmental research. AutoViDev supports Databrary and identifies full-body position estimations in real-time video streams using convolutional pose machine-learning algorithms.
AutoViDev provides valuable information about a variety of behaviors, including gaze direction, facial expressions, posture, locomotion, manual actions, and interactions with objects.
Applying AutoViDev to large-scale, shared video datasets promise to enhance and accelerate research in developmental science.