Constructor and importer of trained serialized models from different dnn-frameworks

net = cv.Net()

net = cv.Net('Caffe', prototxt)
net = cv.Net('Caffe', prototxt, caffeModel)

net = cv.Net('Tensorflow', modelmodel)
net = cv.Net('Tensorflow', model, config)

net = cv.Net('Torch', filename)
net = cv.Net('Torch', filename, isBinary)

net = cv.Net('Darknet', cfgFile)
net = cv.Net('Darknet', cfgFile, darknetModel)

Input

• prototxt path to the .prototxt file with text description of the network architecture.
• caffeModel (optional) path to the .caffemodel file with learned network. Empty by default.
• model path to the .pb file with binary protobuf description of the network architecture. Binary serialized TensorFlow graph includes weights.
• config Optional path to the .pbtxt file that contains text graph definition in protobuf format. Resulting net is built by text graph using weights from a binary one. This is more flexible than binary format and may be used to build the network using binary format only as a weights storage. This approach is similar to Caffe's .prorotxt and .caffemodel.
• filename path to the file, dumped from Torch by using torch.save() function.
• isBinary specifies whether the network was serialized in ascii mode or binary. default true.
• cfgFile path to the .cfg file with text description of the network architecture.
• darknetModel (optional) path to the .weights file with learned network.

The first variant creates an empty network.

The second variant reads a network model stored in Caffe framework's format.

The third variant reads a network model stored in TensorFlow framework's format.

The fourth variant reads a network model stored in Torch7 framework's format.

The fifth variant reads a network model stored in Darknet model files.

The importers first create a net, add loaded layers into it, and set connections between them.

Notes for Torch

NOTE: ASCII mode of Torch serializer is more preferable, because binary mode extensively use long type of C language, which has various bit-length on different systems.

The loading file must contain serialized nn.Module object with importing network. Try to eliminate a custom objects from serialazing data to avoid importing errors.

List of supported layers (i.e. object instances derived from Torch nn.Module class):

• nn.Sequential
• nn.Parallel
• nn.Concat
• nn.Linear
• nn.SpatialConvolution
• nn.SpatialMaxPooling, nn.SpatialAveragePooling
• nn.ReLU, nn.TanH, nn.Sigmoid
• nn.Reshape
• nn.SoftMax, nn.LogSoftMax

Also some equivalents of these classes from cunn, cudnn, and fbcunn may be successfully imported.