DeCAF: A Deep Convolutional Activation Feature for Generic Visual Recognition
We evaluate whether features extracted from the activation of a deep convolutional network trained in a fully supervised fashion on a large, fixed set of object recognition task...