ArcFace: Additive Angular Margin Loss for Deep Face Recognition
Citations
40 citations
39 citations
39 citations
39 citations
Cites methods from "ArcFace: Additive Angular Margin Lo..."
...Both PC and ArcFace losses are designed for improving classification performance on hard samples....
[...]
...Motivated by this, we use an in-house developed loss function, i.e., Probabilistically Compact (PC) loss, for training the MS model and compared with ArcFace [19], the additive angular margin loss for deep face recognition, using the classical softmax loss as the baseline....
[...]
...and compared with ArcFace [19], the additive angular margin loss for deep face recognition, using the classical softmax loss as the baseline....
[...]
39 citations
Cites methods from "ArcFace: Additive Angular Margin Lo..."
...To resolve this, some previous works have used fixed scale parameters [31, 32] to increase the dynamic range of the input to the softmax layer....
[...]
References
123,388 citations
33,597 citations
30,843 citations
13,268 citations
10,447 citations