Inspiration

Infrared and thermal imaging systems are being used more and more in places where regular cameras do not work well such as when it's dark or when people are trying to find someone in a disaster. These systems can get important information but it is hard to understand the pictures of faces they take and it usually takes a long time to do it by hand.

We wanted to see if we could use computer vision and generative AI to turn the pictures from these systems into information that people can understand easily. Of just finding a face we wanted to build a system that could get important details from the picture and make a report that people can use to make decisions.

What FaceIR does

FaceIR is a system that uses AI to analyze pictures of faces taken with cameras. It works in places where it's dark or hard to see.

The system can do the following things automatically:

Find faces in the pictures
Get information about the face like what it looks like
Identify the shape of the face and what is on it
Make a report that's easy to understand
Write a summary of the report in language using generative AI

This means that people do not have to look at the pictures by hand and write down what they see. FaceIR can do it in a few seconds.

How we built FaceIR

FaceIR uses different AI technologies to work.

The computer vision part of the system finds faces in the pictures. Gets information from them. We use different models to get details from the face.

Then we use a language model to turn the information into a report that is easy to understand.

The system is on a website that people can use to upload pictures look at the information that the system found and make reports in time.

Our system uses the following things:

Computer Vision
Deep Learning
Facial Attribute Analysis
Infrared Image Processing
Generative AI
Automated Report Generation

Challenges we faced

Infrared pictures are different from pictures and this makes it hard to use them.

Many systems that analyze faces are made to work with pictures and they do not work well with infrared pictures. We had to change our system to work with these pictures.

Another hard thing was making the system write reports that're easy to understand. The system had to make the reports clear and not too long.

It was also hard to make all the different parts of the system work together. We had to work to make the system easy to use.

Things we are proud of

We are proud of the following things:

Building a system that can analyze infrared pictures of faces
Using computer vision. Generative AI together
Making the system write reports automatically
Making a website that's easy to use
Showing that AI can be used in places where it's dark or hard to see

What we learned

We learned that it is hard to work with pictures that're not like regular pictures. We also learned that using different AI technologies together can help solve real problems.

We got experience with pictures using many different AI systems together and making reports with AI. We also learned how generative AI can make information into something that is more useful for people.