4 ms·
I'm surprised this works well. If the method is a state-of-the-art convolutional neural network architecture, you generally would need more than 10 images -- ev
by therajiv 9y ago
I'm surprised this works well. If the method is a state-of-the-art convolutional neural network architecture, you generally would need more than 10 images -- even with transfer learning -- for passable accuracy. Algorithms for medical diagnosis, for example, generally require between 100 and 200 images to do well. Though those are generally transfer learned from ImageNet CNNs. So I'm curious as to which dataset this face recognition uses for weight initialization, or it uses another ML method entirely.
- drpgq 9y agoWell before deep learning you could get good performance on people in your training set. Source worked as research scientist for a face recognition company since 2003.
- daveguy 9y agoPre deep learning -- using facial feature geometry?
- mikeyanderson 9y agoWhat was the use case? Was it for determining faces in general or distinguishing between different faces?
- lightbyte 9y agoDigging through their blog post, it looks like they are using a model based on OpenFace (https://cmusatyalab.github.io/openface/ https://cmusatyalab.github.io/openface/)