Merge pull request BVLC#2086 from longjon/python-net-spec

Python net specification

examples/python_nets/caffenet.py

@@ -0,0 +1,54 @@
+from caffe import layers as L, params as P, to_proto
+from caffe.proto import caffe_pb2
+
+# helper function for common structures
+
+def conv_relu(bottom, ks, nout, stride=1, pad=0, group=1):
+    conv = L.Convolution(bottom, kernel_size=ks, stride=stride,
+                                num_output=nout, pad=pad, group=group)
+    return conv, L.ReLU(conv, in_place=True)
+
+def fc_relu(bottom, nout):
+    fc = L.InnerProduct(bottom, num_output=nout)
+    return fc, L.ReLU(fc, in_place=True)
+
+def max_pool(bottom, ks, stride=1):
+    return L.Pooling(bottom, pool=P.Pooling.MAX, kernel_size=ks, stride=stride)
+
+def alexnet(lmdb, batch_size=256, include_acc=False):
+    data, label = L.Data(source=lmdb, backend=P.Data.LMDB, batch_size=batch_size, ntop=2,
+        transform_param=dict(crop_size=227, mean_value=[104, 117, 123], mirror=True))
+
+    # the net itself
+    conv1, relu1 = conv_relu(data, 11, 96, stride=4)
+    pool1 = max_pool(relu1, 3, stride=2)
+    norm1 = L.LRN(pool1, local_size=5, alpha=1e-4, beta=0.75)
+    conv2, relu2 = conv_relu(norm1, 5, 256, pad=2, group=2)
+    pool2 = max_pool(relu2, 3, stride=2)
+    norm2 = L.LRN(pool2, local_size=5, alpha=1e-4, beta=0.75)
+    conv3, relu3 = conv_relu(norm2, 3, 384, pad=1)
+    conv4, relu4 = conv_relu(relu3, 3, 384, pad=1, group=2)
+    conv5, relu5 = conv_relu(relu4, 3, 256, pad=1, group=2)
+    pool5 = max_pool(relu5, 3, stride=2)
+    fc6, relu6 = fc_relu(pool5, 4096)
+    drop6 = L.Dropout(relu6, in_place=True)
+    fc7, relu7 = fc_relu(drop6, 4096)
+    drop7 = L.Dropout(relu7, in_place=True)
+    fc8 = L.InnerProduct(drop7, num_output=1000)
+    loss = L.SoftmaxWithLoss(fc8, label)
+
+    if include_acc:
+        acc = L.Accuracy(fc8, label)
+        return to_proto((loss, acc), {v: k for k, v in locals().iteritems()})
+    else:
+        return to_proto(loss, {v: k for k, v in locals().iteritems()})
+
+def make_net():
+    with open('train.prototxt', 'w') as f:
+        print >>f, alexnet('/path/to/caffe-train-lmdb')
+
+    with open('test.prototxt', 'w') as f:
+        print >>f, alexnet('/path/to/caffe-val-lmdb', batch_size=50, include_acc=True)
+
+if __name__ == '__main__':
+    make_net()

python/caffe/layers.py

@@ -0,0 +1,95 @@
+from collections import OrderedDict
+import re
+
+from .proto import caffe_pb2
+from google import protobuf
+
+def uncamel(s):
+    """Convert CamelCase to underscore_case."""
+
+    return re.sub('(?!^)([A-Z])(?=[^A-Z])', r'_\1', s).lower()
+
+def assign_proto(proto, name, val):
+    if isinstance(val, list):
+        getattr(proto, name).extend(val)
+    elif isinstance(val, dict):
+        for k, v in val.iteritems():
+            assign_proto(getattr(proto, name), k, v)
+    else:
+        setattr(proto, name, val)
+
+def to_proto(tops, names):
+    if not isinstance(tops, tuple):
+        tops = (tops,)
+    layers = OrderedDict()
+    for top in tops:
+        top.fn._to_proto(layers, names)
+
+    net = caffe_pb2.NetParameter()
+    net.layer.extend(layers.values())
+    return net
+
+class Top:
+    def __init__(self, fn, n):
+        self.fn = fn
+        self.n = n
+
+class Function:
+    def __init__(self, type_name, inputs, params):
+        self.type_name = type_name
+        self.inputs = inputs
+        self.params = params
+        self.ntop = self.params.get('ntop', 1)
+        if 'ntop' in self.params:
+            del self.params['ntop']
+        self.in_place = self.params.get('in_place', False)
+        if 'in_place' in self.params:
+            del self.params['in_place']
+        self.tops = tuple(Top(self, n) for n in range(self.ntop))
+
+    def _to_proto(self, layers, names):
+        bottom_names = []
+        for inp in self.inputs:
+            if inp.fn not in layers:
+                inp.fn._to_proto(layers, names)
+            bottom_names.append(layers[inp.fn].top[inp.n])
+        layer = caffe_pb2.LayerParameter()
+        layer.type = self.type_name
+        layer.bottom.extend(bottom_names)
+
+        if self.in_place:
+            layer.top.extend(layer.bottom)
+            layer.name = names[self.tops[0]]
+        else:
+            for top in self.tops:
+                layer.top.append(names[top])
+            layer.name = layer.top[0]
+
+        for k, v in self.params.iteritems():
+            # special case to handle generic *params
+            if k.endswith('param'):
+                assign_proto(layer, k, v)
+            else:
+                assign_proto(getattr(layer, uncamel(self.type_name) + '_param'), k, v)
+
+        layers[self] = layer
+
+class Layers:
+    def __getattr__(self, name):
+        def layer_fn(*args, **kwargs):
+            fn = Function(name, args, kwargs)
+            if fn.ntop == 1:
+                return fn.tops[0]
+            else:
+                return fn.tops
+        return layer_fn
+
+class Parameters:
+    def __getattr__(self, name):
+       class Param:
+            def __getattr__(self, param_name):
+                return getattr(getattr(caffe_pb2, name + 'Parameter'), param_name)
+       return Param()
+
+layers = Layers()
+params = Parameters()