Closed lynnw123 closed 5 years ago
To my first question, I think to vote the max number as the age prediction is a better way. I am wondering is there a way that I could add some softmax above certain attributes, like group softmax in PyTorch?
@YL-123 I treat the attribute classification as a multi-label classification problem, which is a simple way. As you said, different attributes may be conflict with each other, e.g. age16-30 and age13-45, thus using softmax on these mutually exclusive attributes would be better where the ambiguity would be decreased.
Hello, thanks for sharing your great work! Could you help me understand the following sentence in the paper, "the DeepMAR output a label vector which represents whether it has each attribute or not"? For example, in the case of the age prediction, PETA dataset which had four attributes 16-30, 31-45, 45-60 and >60 in your selected 35 attributes, If more than 1 output has a positive score, (eg. first twos) it means the image has both 16-30 and 31-45 attribute? or we need sort above four output scores and vote the biggest positive number as the age prediction? I noticed one more attribute <16 later in the PETA datasets. have you performed some experiments about this? thanks again!