Skip to main content


woman holding baby, both looking at camera


The MIT Quest for Intelligence is launching an updated  version of iCatcher+, a tool to automate annotation of video data from studies of infants and children.

Many neurological and psychological studies are interested in how babies see, learn, and think, by showing them stimuli and then measuring where they look, and for how long. Gathering that data is difficult — the children are squirmy, the experiments can be long, and it takes many researcher hours to watch and annotate video recordings of each study. As part of a multi-institution collaboration, the Engineering Team at the MIT Quest for Intelligence have developed a new version of iCatcher+, a machine learning tool that automates this annotation at near-human accuracy, greatly increasing the speed and ease at which this video data can be analyzed and understood.

During the pandemic, many cognitive scientists began to gather research data through synchronous video meetings, like Zoom, or asynchronous platforms like Lookit, rather than in person. What started as a safety measure that allowed research to continue during lockdown has resulted in access to much wider pool of participants. A few years ago, only families living near a research facility were able to come in-person to a research session, and it may have required giving up a day of work and making a special trip with a small child or infant. Now, families can participate from home, with minimal time commitment and no expense for travel, which means that researchers have access to a much larger pool of potential participants from different geographic areas and socio-economic backgrounds, which increases the speed and generalizability of the research. However, the increased number of study recordings are still slow to analyze, because labs often rely on humans to do these tasks. iCatcher+ addresses this bottleneck by analyzing the recordings more quickly than humans can (and at near-human accuracy), and adding a user-friendly interface for researchers to easily review the analyzed data. It’s a remarkable breakthrough for cognitive scientists focused on early childhood development — and one of the researchers says that it is a win for democracy in how it has expanded the data that can be analyzed.

iCatcher+ was developed via multi-institution collaboration involving significant effort and leadership across MIT.

Back to Projects