Sifter T-shirt identifier
Technologies used: Android, OpenCV, C, C++, Docker
Written as a side project during my time as a SWE at Threadless, Sifter leverages computer vision techniques to do realtime design identification of Threadless t-shirt designs, from a database of over 3,000 designs, with a photo captured with an Android phone.
When I work for a company, I cultivate a passion for what they do. Sifter was a fun side project to express the passion that I had for the artists of Threadless. With so many popular T-shirt designs laying around the office, it was impossible to remember all of their names, and so the idea to identify them with Sifter was born!
I built the full stack, from the fullscreen camera Android app which captures to the image, to the web API running in a docker container on AWS, to the C++ OpenCV-based backend system which performs the fast feature matching and searching.