Extracting feature vectors from inception_v3 model and creating feature embedding

Hi,

I have downloaded inception_v3 pretrained model from https://s3.amazonaws.com/lasagne/recipes/pretrained/imagenet/inception_v3.pkl.

By using this model, I want to extract image feature vectors like in tensorflow (len=2048), and create image embeddings (len=100). How is it possible?

Is there any tutorial or example about it?

Thanks,