Child In Crowd
As a parent and developer I couldn't believe that there wasn't a technological solution for relating parents and children in crowded events. It is true that there are other solutions like invisible stamps that match between children and parents or printed tickets, but I believe that artificial intelligence can bring a cheaper and more comfortable solution. For this reason Child in Crowd was developed in order to bring to the society an open software for relating children and parents in crowds.
What it does
Child in Crowd extracts the biometrics data from a picture identifying infants and relating them with the adults that appear in the image; this image is considered the check-in operation. As a second image is taken with the face of the same child the program checks if at least one adult in this second picture corresponds with one of the adults of the check-in image. If none of the adults corresponds, a warning message is displayed and you are prompted to see the check-in image right from the interface.
How I built it
Child in Crowd is developed using NodeJS and Vue.js. The backend is based on AWS lambda functions and the faces recognition is performed using AWS artificial intelligence services.
Challenges I ran into
The first challenge was to define the logic in which registration and check-out is performed in order to assure a frictionless experience. The second challenge was to fulfill the AWS Serverless Applications Repository requirements for the software permissions in the cloud.
Accomplishments that I'm proud of
I am proud of the functional coming approach that let users register multiple children from a single image.
What I learned
I learned a lot about programmatic image processing for cropping and normalizing.
What's next for Child In Crowd
I was deeply inspired by this project that now i would like to implement technology based non-profit organization for children support against predators.
Online interface screenshots
Child in Crowd warning on non-corresponding relative
Child in Crowd successful checkout message