This is intended to serve as a conversion module for Caffe models to Keras Functional API models.
Please, be aware that this feature is not regularly maintained. Thus, some layers or parameter definitions introduced in newer versions of either Keras or Caffe might not be compatible with the converter.
For this reason, any pull requests with updated versions of the caffe2keras converter are highly welcome!
In order to convert a model you just need the .caffemodel weights and the .prototxt deploy or train file. In any case you will need to include the input image dimensions as a header to the .prototxt network structure as if it was a deploy model (see an example here) and also include an initial data layer:
layer {
name: "data"
type: "Data"
top: "data"
top: "label"
}
Given the differences between Caffe and Keras when applying the MAX pooling opperation, in some occasions the MAX pooling layers must include a pad: 1 value even if they did not include them in their original .prototxt.
The file caffe2keras.py can be used as a command line interface for converting any model the following way:
python caffe2keras.py -load_path 'models/' -prototxt 'train_val_for_keras.prototxt' -caffemodel 'bvlc_googlenet.caffemodel'
In the file test_converted.py you can see an example on how to use a converted model.
This code is a modified and improved version by Marc Bolaños of the original pieces of code originally written by Pranav Shyam and Antonella Cascitelli.
Contact email: marc.bolanos@ub.edu GitHub page: https://github.com/MarcBS