This is a proof-of-concept that shows how to use pretrained Caffe models in TensorFlow.

This demo imports the VGG-16 network network.

Verified on the ILSVRC2012 validation set, on which it achieves a top 5 accuracy of 86.64%.

A minor quirk: you need to run kaffe.py to extract the parameters first, since it appears that Caffe and TensorFlow don't play nice (CUDA conflicts). This persists even with set_mode_cpu. However, for future work, there's really no need to depend on Caffe here: just the protocol buffers should suffice.