Did you re-train the googleNet for image retrieval? How did you compress the output of googlenet into 96 dimension vector?
Did you re-train the googleNet for image retrieval? How did you compress the output of googlenet into 96 dimension vector?