As far as I could understand, you converted the multi-class classification into independent binary classification, so I assume in Eq. (4) $y_i$ should be the label for one class and therefore, there should be a summation over all classes? I mean it should look like:
where $C$ is the total number of classes in training.
Hi @ZiqinZhou66 ,
Thanks for your great work ZegCLIP. However, I am a bit confused about Equation (4) in your paper https://arxiv.org/pdf/2212.03588.pdf
As far as I could understand, you converted the multi-class classification into independent binary classification, so I assume in Eq. (4) $y_i$ should be the label for one class and therefore, there should be a summation over all classes? I mean it should look like:
where $C$ is the total number of classes in training.
Looking forward to your reply.
Many thanks, Yiming