Mobile Image Blur Detection with Machine Learning

snapADDY is now also on Medium! On the new snapADDY Tech Blog our development team will give you detailed technical insights into the further development and daily work at snapADDY.

In a recent post on the blogging platform Medium, our Chief Data Officer, Dr. Benedikt Brief, gives insights into our data team's current development work at snapADDY

The article is about improving the user experience when scanning business cards: combining techniques from image processing with algorithms from Machine Learning, snapADDY products will be able to immediately detect blurry or unsteady photos. Users will be notified via an improved camera view and can quickly retake the photo, if necessary.

From a technical perspective, the article shows the complete process, from problem statement and collecting test data, to developing a prototype in Python, to implementing the system in the final camera view in TypeScript. Along the way, you will learn about some best practices and experiences from this project.