Infant Pose

Infant pose estimation, the automated analysis of an infant's body posture from images or videos, is a rapidly developing field driven by the need for objective, efficient assessment of motor development and early detection of developmental disorders. Current research focuses on improving the accuracy of pose estimation models, particularly addressing challenges posed by the unique characteristics of infant bodies and the limited availability of annotated infant data, often employing techniques like domain adaptation, generative models, and transformer-based architectures. These advancements have implications for automating clinical assessments like the General Movement Assessment, enabling earlier diagnosis and intervention for conditions such as autism spectrum disorder and congenital muscular torticollis, and facilitating non-contact monitoring in healthcare and home settings.