I just tested the library with model GoogleNet, which should be the same model of Tensorflow demo and it runs more than 10s for each image, which is much more than the TF demo.
Is it because the caffe model doesn't scale the image size while the TensorFlow demo scales it to a relatively smaller one (224 * 224). Or there are some other things I've missed?
I just tested the library with model GoogleNet, which should be the same model of Tensorflow demo and it runs more than 10s for each image, which is much more than the TF demo. Is it because the caffe model doesn't scale the image size while the TensorFlow demo scales it to a relatively smaller one (224 * 224). Or there are some other things I've missed?