Beginning in infancy, individuals acquire the social and communication skills that are vital for a healthy and productive life. Children with developmental delays face great challenges in acquiring these skills, resulting in substantial lifetime risks. Children with an Autism Spectrum Disorder (ASD) represent a particularly significant risk category, due both to the increasing rate of diagnosis of ASD and its consequences. Since the genetic basis for ASD is unclear, the diagnosis, treatment, and study of the disorder depends fundamentally on the observation of behavior. Unfortunately, current methods for acquiring and analyzing behavioral data are so labor-intensive as to preclude their large scale application. In this talk, I will describe our research agenda in Behavior Imaging, which targets the capture, modeling, and analysis of social and communicative behaviors between children and their caregivers and peers. We are developing computational methods and statistical models for the analysis of vision, audio, and wearable sensor data. Our goal is to develop a new set of capabilities for the large-scale collection and interpretation of behavioral data. I will describe several research challenges in statistical modeling which arise in this area, and present illustrative results in the video-based analysis of social interactions.