support lenet and resnet convertion

fluid/image_classification/caffe2fluid/README.md

 ### caffe2fluid
     this tool is used to convert a caffe-model to paddle-model(fluid api)
 
-
 ### howto
-    0, prepare caffepb.py in ./proto
+    1, prepare caffepb.py in ./proto, two options provided
         option 1: generate it from caffe.proto using protoc
             bash ./proto/compile.sh
 
         option2: download one from github directly
             cd proto/ && wget https://github.com/ethereon/caffe-tensorflow/blob/master/kaffe/caffe/caffepb.py
 
-    1, convert you caffe model using convert.py which will generate a python code and weight(in .npy)
+    2, convert you caffe model using convert.py which will generate a python code and weight(in .npy)
 
-    2, use the converted model to predict
+    3, use the converted model to predict
 
     (see more detail info in 'tests/lenet/README.md')
 

fluid/image_classification/caffe2fluid/kaffe/caffe/caffepb.py

-../../proto/caffepb.py
\ No newline at end of file

fluid/image_classification/caffe2fluid/kaffe/caffe/resolver.py

+import os
 import sys
 
 SHARED_CAFFE_RESOLVER = None
 
 
+def import_caffepb():
+    p = os.path.realpath(__file__)
+    p = os.path.dirname(p)
+    p = os.path.join(p, '../../proto')
+    sys.path.insert(0, p)
+    import caffepb
+    return caffepb
+
+
 class CaffeResolver(object):
     def __init__(self):
         self.import_caffe()
@@ -15,8 +25,7 @@ class CaffeResolver(object):
             self.caffe = caffe
         except ImportError:
             # Fall back to the protobuf implementation
-            from . import caffepb
-            self.caffepb = caffepb
+            self.caffepb = import_caffepb()
             show_fallback_warning()
         if self.caffe:
             # Use the protobuf code from the imported distribution.

fluid/image_classification/caffe2fluid/kaffe/paddle/network.py

@@ -165,7 +165,6 @@ class Network(object):
 
         # Get the number of channels in the input
         h_i, w_i = input.shape[2:]
-
         fluid = import_fluid()
         output = fluid.layers.pool2d(
             input=input,
@@ -182,7 +181,6 @@ class Network(object):
 
         # Get the number of channels in the input
         h_i, w_i = input.shape[2:]
-
         fluid = import_fluid()
         output = fluid.layers.pool2d(
             input=input,

fluid/image_classification/caffe2fluid/kaffe/paddle/transformer.py

@@ -273,7 +273,6 @@ class TensorFlowEmitter(object):
                 b += self.emit_node(node)
             blocks.append(b[:-1])
         s = s + '\n\n'.join(blocks)
-
         s += self.emit_convert_def(input_nodes)
         s += self.emit_main_def(name)
         return s

fluid/image_classification/caffe2fluid/tests/lenet/README.md

 ### convert lenet model from caffe format into paddle format(fluid api)
 
 ### howto
+    0, prepare your caffepb.py
+
     1, download a lenet caffe-model
         lenet_iter_10000.caffemodel
             download address: https://github.com/ethereon/caffe-tensorflow/raw/master/examples/mnist/lenet_iter_10000.caffemodel

fluid/image_classification/caffe2fluid/tests/lenet/convert.sh

@@ -2,6 +2,8 @@
 
 #function:
 #   convert a caffe model
+#   eg:
+#       bash ./convert.sh ./model.caffe/lenet.prototxt ./model.caffe/lenet.caffemodel lenet.py lenet.npy
 
 if [[ $# -ne 4 ]];then
     echo "usage:"

fluid/image_classification/caffe2fluid/tests/lenet/lenet.py

@@ -61,8 +61,8 @@ class Network(object):
         fluid = import_fluid()
         #load fluid mode directly
         if os.path.isdir(data_path):
-            assert (
-                exe is not None), 'must provide a executor to load fluid model'
+            assert (exe is not None), \
+                'must provide a executor to load fluid model'
             fluid.io.load_persistables_if_exist(executor=exe, dirname=data_path)
             return True
 
@@ -167,7 +167,6 @@ class Network(object):
 
         # Get the number of channels in the input
         h_i, w_i = input.shape[2:]
-
         fluid = import_fluid()
         output = fluid.layers.pool2d(
             input=input,
@@ -184,7 +183,6 @@ class Network(object):
 
         # Get the number of channels in the input
         h_i, w_i = input.shape[2:]
-
         fluid = import_fluid()
         output = fluid.layers.pool2d(
             input=input,