CompreFace – the Best Open-Source Facial Recognition Software?

Share article

Best Tool

A success story about how our face recognition solution — CompreFace — and how it benefits people in the wider developer community.

Today’s guests are Drake, a rising junior at Wabash College in Indiana, and Selena from the University of Portland in Portland, Oregon. They’re working at the REU program through the University of Houston’s Computer Science Department. In search of the best open-source facial recognition project, they chose CompreFace and are ready to share the results.

Now for the fun part!

  1. What project are you working on and why did you choose CompreFace for it?

    Our project involves using facial recognition software in a similar way to how one of our professors has used fingerprints. He used a fingerprint database and a scanner to determine if sample prints were real or not. We’re doing the same thing with faces, and needed some accurate facial recognition software to get the job done optimally.

    The NIST catalog was recommended to us, and we found a lot of software in it. After a significant amount of focused research and coming across an article that ranked CompreFace as #1, we decided to give it a go. First, we checked your github and downloaded it. It was much easier to get on our MacBook than the software we were previously using, which we had had some issues with. We also noticed that CompreFace’s description on github matched perfectly with what we were aiming to achieve.

    As part of our project, we used open-source photo editing software to distort images of features, including the width and height of an image, making the faces look different. We then ran these images through CompreFace to see if these images were still recognized as faces.

  2. Can you please share the results of your research?

    Not really, I’m afraid! We’re still very much in the process but it’s making good progress. I’ll say that. However, we’ll be happy to share the final results once we’re done with the project.

  3. Did you try other face recognition systems?

    Yes, we tried Betaface before we tested CompreFace. We had some issues with licensing with them. We looked into several other systems, but we couldn’t download some of them, others weren’t compatible with MacBooks, and some had confusing githubs. We reached out to everyone via email, but only your team got back to us with sufficient details. Throughout the whole process, CompreFace has stood out as being super useful on multiple levels.

We’ll be following up with Selena and Drake to see their project results. If you’d like to share a similar story of how CompreFace helped you in your project or research, contact our team.

Ready to chat? Join CompreFace community!

Join us! Exadel provides cutting-edge products and services. Check out our open positions here.