With both papers released at around the same time, neither results compared with each other. Looking at the validation logs on Vggface2, it seems both performed similarly at 80k iterations. Hence, what are the difference between both methods and are there any situations where 1 is better than the other?
Extra question: is there any example code on how to use Sphereface to do own recognition instead of the dataset?
With both papers released at around the same time, neither results compared with each other. Looking at the validation logs on Vggface2, it seems both performed similarly at 80k iterations. Hence, what are the difference between both methods and are there any situations where 1 is better than the other?
Extra question: is there any example code on how to use Sphereface to do own recognition instead of the dataset?