Identifying cat faces in a picture is nothing new. Back in 2012, Google uses 1000 machines to implement one of the earliest deep learning network to learn the concept of cat faces. However it becomes challenging when you try to use it as identification in a passive setup where cat keeps moving around while camera remains in a fixed position with limited FOV. In this talk we would like share the experience when we were building CatFi to make cat identification possible and reliable.
Mu Chi is the founder and CEO of CatFi, which is the world's smartest cat feeder with novel cat facial recognition to track the health status for each individual cats. He has been involved in GPU software development since 2006 and started Zillians to foster GPU-accelerated cloud in 2008.