python - How to predict float vector labels with caffe? -
i wondering if it's possible predict 1-by-n feature associated input image using caffe. in this post there solution make caffe predict binary vector such [1 0 1 0].
is solution suitable if have 1-by-n float vector label (such [0.2, 0.1, 0.3, 0.4] ? want predict such vector, not binary vector label.
you can think multitaskdata layer. can parse float typed label vector mentioned in question.
Comments
Post a Comment