You have a good point, but there are better ways to put it—e.g. ask if such data is available, or how the author tests the work (this is a Show HN, the author's here)—than a swipe like "next to useless".
For example, check out our YouTube video of a demo training a classifier in real-time with just 10 images per person at https://www.youtube.com/watch?v=LZJOTRkjZA4. This demo is included in the repo and the README has instructions on running it.
Also note that there is a distinction between training the neural network, which extracts the face representations, from using the features for tasks like clustering and classification.
How different is OpenFace vs. OpenBR?
--
As our initial ROC curve on LFW's similarity benchmark in https://github.com/cmusatyalab/openface/blob/master/images/n... shows, this approach results in slightly improved performance. The best point is an FPR of 0.0 and TPR of 1.0 (top left). You can see today's state-of-the-art private systems in the top left, followed by open source systems, then by historical techniques OpenCV provides like Eigenfaces. The dashed line in the middle shows what randomly guessing would provide.
OpenBR is going in a great direction for reproducible and open face recognition. They provide a pipeline for preprocessing and representing faces, as well as doing similarity and classification tasks on the representations. The techniques from OpenFace could be integrated into OpenBR's pipeline.
Training new models is currently dominated by huge industry datasets, which currently have 100's of millions of images. My current dataset is from datasets available for research and has ~500k images.
