using pycaffe to read prototxt

[Ali Mousavi]

I'm trying to read my caffe models in python so I can take the last train caffemodel, add some weights to it and use it as an initialization for another file.

Im new to pycaffe and I get the following error:

    libprotobuf ERROR google/protobuf/text_format.cc:172] Error parsing text-format caffe.NetParameter: 1:10: Message type "caffe.NetParameter" has no field named "train_net".

    WARNING: Logging before InitGoogleLogging() is written to STDERR

    F0831 20:19:51.970383 15975 upgrade_proto.cpp:928] Check failed: ReadProtoFromTextFile(param_file, param) Failed to parse NetParameter file: ./voc12/config/vgg128_noup/solver_train_aug.prototxt

    *** Check failure stack trace: ***

    Aborted (core dumped)

here is my python code

    import numpy as np

    import Image

    caffe_root = './'  # this file is expected to be in {caffe_root}/examples

    import sys

    sys.path.insert(0, caffe_root + 'python')

    

    import caffe

    

    

    # Load the original network and extract the fully connected layers' parameters.

    model= './voc12/config/vgg128_noup/solver_train_aug.prototxt'

    #model = './examples/mnist/lenet_solver.prototxt'

    weights = './voc12/model/vgg128_noup/train_iter_6000.caffemodel'

    net = caffe.Classifier(model, 

                    weights)

below is my solver prototext:

    train_net: "voc12/config/vgg128_noup/train_train_aug.prototxt"

    

    lr_policy: "step"

    gamma: 0.1

    stepsize: 2000

    base_lr: 0.001

    

    display: 10

    max_iter: 6000

    momentum: 0.9

    weight_decay: 0.0005

    

    snapshot: 2000

    snapshot_prefix: "voc12/model/vgg128_noup/train"

    solver_mode: GPU

                                                                                                                                                                                    

any suggestions                   

[Cyndi]

Your model should be the your train/test deploy prototxt, not the solver prototxt (for example, look at "models/bvlc_reference_caffenet/deploy.prototxt")