質問編集履歴

情報の追加

2017/04/11 11:42

投稿

退会済みユーザー

スコア0

test CHANGED Viewed

File without changes

test CHANGED Viewed

@@ -4,6 +4,136 @@
 ```ここに言語を入力
+# -*- coding: utf-8 -*-
+from __future__ import absolute_import
+from __future__ import division
+from __future__ import print_function
+import os
+import numpy as np
+class Cifar10Record(object):
+    width = 32
+    height = 32
+    depth = 3
+    def set_label(self,label_byte):
+        #self.label = np.frombuffer(label_byte,dtype=np.unit8)
+        self.label = np.frombuffer(label_byte,dtype=np.uint8) # unit8 -> uint8に修正
+    def set_image(self,image_bytes):
+        byte_buffer = np.frombuffer(image_bytes,dtype=np.int8)
+        reshaped_array = np.reshape(byte_buffer,[self.depth,self.height,self.width])
+        self.byte_array = np.transpose(reshaped_array,[1,2,0])
+        self.byte_array = self.byte_array.astype(np.float32)
+class Cifar10Reader(object):
+ def __init__(self,filename):
+     if not os.path.exists(filename):
+         print(filename + ' is not exist')
+         return
+     self.bytestream = open(filename,mode="rb")
+ def close(self):
+     if not self.bytestream:
+         self.bytestream.close()
+ def read(self,index):
+     result = Cifar10Record()
+     label_bytes = 1
+     image_bytes = result.height * result.width * result.depth
+     record_bytes = label_bytes + image_bytes
+     self.bytestream.seek(record_bytes * index,0)
+     result.set_label(self.bytestream.read(label_bytes))
+     result.set_image(self.bytestream.read(image_bytes))
+     return result
+     print(self.bytestream)
+# 追加
+reader = Cifar10Reader("lena_std.tif")
+ret = reader.read(0)
+print(ret)
+```
+というCIFAR-10形式のデータセットを読み込むプログラム を書きました。
+参考url:
+http://www.buildinsider.net/small/booktensorflow/0201
+他に使用しているスクリプトは
+png10.py
+```ここに言語を入力
 # coding: utf-8
@@ -18,103 +148,71 @@
 import os
 import numpy as np
-class Cifar10Record(object):
-    width = 32
-    height = 32
-    depth = 3
-    def set_label(self,label_byte):
-        self.label = np.frombuffer(label_byte,dtype=np.unit8)
-    def set_image(self,image_bytes):
-        byte_buffer = np.frombuffer(image_bytes,dtype=np.int8)
-        reshaped_array = np.reshape(byte_buffer,[self.depth,self.height,self.width])
-        self.byte_array = np.transpose(reshaped_array,[1,2,0])
-        self.byte_array = self.byte_array.astype(np.float32)
-class Cifar10Reader(object):
- def __init__(self,filename):
-     if not os.path.exists(filename):
-         print(filename + ' is not exist')
-         return
-     self.bytestream = open(filename,mode="rb")
- def close(self):
-     if not self.bytestream:
-         self.bytestream.close()
- def read(self,index):
-     result = Cifar10Record()
-     label_bytes = 1
-     image_bytes = result.height * result.width * result.depth
-     record_bytes = label_bytes + image_bytes
-     self.bytestream.seek(record_bytes * index,0)
-     result.set_label(self.bytestream.read(label_bytes))
-     result.set_image(self.bytestream.read(image_bytes))
-     return result
+import tensorflow as tf
+from PIL import Image
+from reader import Cifar10Reader
+FLAGS = tf.app.flags.FLAGS
+tf.app.flags.DEFINE_string('file',None,"処理するファイルのパス")
+tf.app.flags.DEFINE_integer('offset',0,"読み飛ばすレコード数")
+tf.app.flags.DEFINE_integer('length',16,"読み込んで変換するレコード数")
+basename = os.path.basename(FLAGS.file)
+path = os.path.dirname(FLAGS.file)
+reader = Cifar10Reader(FLAGS.file)
+stop = FLAGS.offset + FLAGS.length
+for index in range(FLAGS.offset,stop):
+    image = reader.read(index)
+    print('label: %d' % image.label)
+    imageshow = Image.fromarray(image.byte_array.astype(np.unit8))
+    file_name = '%s-%02d-%d.png' % (basename,index,image.label)
+    file = os.path.join(path,file_name)
+    with open(file,mode='wb') as out:
+        imageshow.save(out,format='png')
+reader.close()
 ```
-というCIFAR-10形式のデータセットを読み込むプログラム を書きました。
-参考url:
-http://www.buildinsider.net/small/booktensorflow/0201
-他に使用しているスクリプトは
-png10.py
+model.py
 ```ここに言語を入力
@@ -130,15 +228,169 @@
+import tensorflow as tf
+NUM_CLASSES = 10
+def _get_weights(shape,stddev=1.0):
+  var = tf.get_variable(
+      'weights',
+      shape,
+      initializer=tf.truncated_normal_initializer(stddev=stddev)
+  )
+  return var
+def _get_biases(shape,value=0.0):
+  var = tf.get_variable(
+      'biases',
+      shape,
+      initializer=tf.constant_initializer(value)
+  )
+  return var
+def inference(image_node):
+   #  conv1
+   with tf.variable_scope('conv1') as scope:
+     weights = _get_weights(shape=[5,5,3,64],stddev=1e-4)
+     conv = tf.nn.conv2d(image_node,weights,[1,1,1,1],padding='SAME')
+     biases = _get_biases([64],value=0.1)
+     bias = tf.nn.bias_add(conv,biases)
+     conv1 = tf.nn.relu(bias,name=scope.name)
+   #  pool
+   pool1 = tf.nn.max_pool(conv1,ksize=[1,3,3,1],strides=[1,2,2,1],padding='SAME',name='pool1')
+   #  conv2
+   with tf.variable_scope('conv2') as scope:
+     weights = _get_weights(shape=[5,5,64,64],stddev=1e-4)
+     conv = tf.nn.conv2d(pool1,weights,[1,1,1,1],padding='SAME')
+     biases = _get_biases([64],value=0.1)
+     bias = tf.nn.bias_add(conv,biases)
+     conv2 = tf.nn.relu(bias,name=scope.name)
+   #  pool2
+   pool2 = tf.nn.max_pool(conv2,ksize=[1,3,3,1],strides=[1,2,2,1],padding='SAME',name='pool2')
+   reshape = tf.reshape(pool2,[1,-1])
+   dim = reshape.get_shape()[1].value
+   #  fc3
+   with tf.variable_scope('fc3') as scope:
+       weights = _get_weights(shape=[dim,384],stddev=0.04)
+       biases = _get_biases([384],value=0.1)
+       fc3 = tf.nn.relu(
+           tf.matmul(reshape,weights) + biases,
+           name=scope.name
+       )
+   #  fc4
+   with tf.variable_scope('fc4') as scope:
+       weights = _get_weights(shape=[384,192],stddev=0.04)
+       biases = _get_biases([192],value=0.1)
+       fc4 = tf.nn.relu(tf.matmul(fc3,weights) + biases,name=scope.name)
+   #   output
+   with tf.variable_scope('output') as scope:
+       weights = _get_weights(shape=[192,NUM_CLASSES],stddev=1/192.0)
+       biases = _get_biases([NUM_CLASSES],value=0.0)
+       logits = tf.add(tf.matmul(fc4,weights),biases,name='logits')
+   return logits
+```
+inference.py
+```ここに言語を入力
+# coding: utf-8
+from __future__ import absolute_import
+from __future__ import division
+from __future__ import print_function
 import os
-import numpy as np
+import time
 import tensorflow as tf
-from PIL import Image
+import model as model
@@ -148,358 +400,122 @@
 FLAGS = tf.app.flags.FLAGS
-tf.app.flags.DEFINE_string('file',None,"処理するファイルのパス")
-tf.app.flags.DEFINE_integer('offset',0,"読み飛ばすレコード数")
+tf.app.flags.DEFINE_integer('epoch',30,"訓練するEpoch数")
-tf.app.flags.DEFINE_integer('length',16,"読み込んで変換するレコード数")
+tf.app.flags.DEFINE_string('data_dir','./data/',"訓練データのディレクトリ")
+tf.app.flags.DEFINE_string('checkpoint_dir','./checkpoints/',"チェックポイントを保存するディレクトリ")
+filenames = [
+    os.path.join(FLAGS.data_dir,'data_batch_%d.bin' % i) for i in range(1,6)
+]
+def main(argv=None):
+    train_placeholder = tf.placeholder(tf.float32,shape=[32,32,3],name='input_image')
-basename = os.path.basename(FLAGS.file)
+    image_node = tf.expand_dims(train_placeholder,0)
-path = os.path.dirname(FLAGS.file)
+    logits = model.inference(image_node)
+    with tf.Session() as sess:
+        sess.run(tf.initialize_all_variables())
+        total_duration = 0
+        for epoch in range(1,FLAGS.epoch+1):
+            start_time = time.time()
+            for file_index in range(5):
+                print('Epoch %d: %s' % (epoch,filenames[file_index]))
-reader = Cifar10Reader(FLAGS.file)
+                reader = Cifar10Reader(filenames[file_index])
-stop = FLAGS.offset + FLAGS.length
-for index in range(FLAGS.offset,stop):
+                for index in range(10000):
+                    image = reader.read(index)
+                    logits_value = sess.run([logits],feed_dict={
+                        train_placeholder:image.byte_array,
+                    })
+                    if index % 1000 ==0:
+                        print('[%d]: %r'% (image.label,logits_value))
+            reader.close()
+            duration = time.time() - start_time
+            total_duration += duration
+            print('epoch %d duration = %d sec'%(epoch,duration))
+            tf.train.SummaryWriter(FLAGS.checkpoint_dir,sess.graph)
+        print('Total duration = %d sec'% total_duration)
+if __name__ == '__main__':
+    tf.app.run()
+```
+しかし、inference.py を実行すると
+```ここに言語を入力
+File "inference.py", line 44, in main
     image = reader.read(index)
+  File "/Users/XXX/Desktop/cifar/reader.py", line 43, in read
-    print('label: %d' % image.label)
+    self.bytestream.seek(record_bytes * index,0)
-    imageshow = Image.fromarray(image.byte_array.astype(np.unit8))
-    file_name = '%s-%02d-%d.png' % (basename,index,image.label)
+AttributeError: 'Cifar10Reader' object has no attribute 'bytestream'
-    file = os.path.join(path,file_name)
-    with open(file,mode='wb') as out:
-        imageshow.save(out,format='png')
-reader.close()
 ```
-model.py
-```ここに言語を入力
-# coding: utf-8
-from __future__ import absolute_import
-from __future__ import division
-from __future__ import print_function
-import tensorflow as tf
-NUM_CLASSES = 10
-def _get_weights(shape,stddev=1.0):
-  var = tf.get_variable(
-      'weights',
-      shape,
-      initializer=tf.truncated_normal_initializer(stddev=stddev)
-  )
-  return var
-def _get_biases(shape,value=0.0):
-  var = tf.get_variable(
-      'biases',
-      shape,
-      initializer=tf.constant_initializer(value)
-  )
-  return var
-def inference(image_node):
-   #  conv1
-   with tf.variable_scope('conv1') as scope:
-     weights = _get_weights(shape=[5,5,3,64],stddev=1e-4)
-     conv = tf.nn.conv2d(image_node,weights,[1,1,1,1],padding='SAME')
-     biases = _get_biases([64],value=0.1)
-     bias = tf.nn.bias_add(conv,biases)
-     conv1 = tf.nn.relu(bias,name=scope.name)
-   #  pool
-   pool1 = tf.nn.max_pool(conv1,ksize=[1,3,3,1],strides=[1,2,2,1],padding='SAME',name='pool1')
-   #  conv2
-   with tf.variable_scope('conv2') as scope:
-     weights = _get_weights(shape=[5,5,64,64],stddev=1e-4)
-     conv = tf.nn.conv2d(pool1,weights,[1,1,1,1],padding='SAME')
-     biases = _get_biases([64],value=0.1)
-     bias = tf.nn.bias_add(conv,biases)
-     conv2 = tf.nn.relu(bias,name=scope.name)
-   #  pool2
-   pool2 = tf.nn.max_pool(conv2,ksize=[1,3,3,1],strides=[1,2,2,1],padding='SAME',name='pool2')
-   reshape = tf.reshape(pool2,[1,-1])
-   dim = reshape.get_shape()[1].value
-   #  fc3
-   with tf.variable_scope('fc3') as scope:
-       weights = _get_weights(shape=[dim,384],stddev=0.04)
-       biases = _get_biases([384],value=0.1)
-       fc3 = tf.nn.relu(
-           tf.matmul(reshape,weights) + biases,
-           name=scope.name
-       )
-   #  fc4
-   with tf.variable_scope('fc4') as scope:
-       weights = _get_weights(shape=[384,192],stddev=0.04)
-       biases = _get_biases([192],value=0.1)
-       fc4 = tf.nn.relu(tf.matmul(fc3,weights) + biases,name=scope.name)
-   #   output
-   with tf.variable_scope('output') as scope:
-       weights = _get_weights(shape=[192,NUM_CLASSES],stddev=1/192.0)
-       biases = _get_biases([NUM_CLASSES],value=0.0)
-       logits = tf.add(tf.matmul(fc4,weights),biases,name='logits')
-   return logits
-```
-inference.py
-```ここに言語を入力
-# coding: utf-8
-from __future__ import absolute_import
-from __future__ import division
-from __future__ import print_function
-import os
-import time
-import tensorflow as tf
-import model as model
-from reader import Cifar10Reader
-FLAGS = tf.app.flags.FLAGS
-tf.app.flags.DEFINE_integer('epoch',30,"訓練するEpoch数")
-tf.app.flags.DEFINE_string('data_dir','./data/',"訓練データのディレクトリ")
-tf.app.flags.DEFINE_string('checkpoint_dir','./checkpoints/',"チェックポイントを保存するディレクトリ")
-filenames = [
-    os.path.join(FLAGS.data_dir,'data_batch_%d.bin' % i) for i in range(1,6)
-]
-def main(argv=None):
-    train_placeholder = tf.placeholder(tf.float32,shape=[32,32,3],name='input_image')
-    image_node = tf.expand_dims(train_placeholder,0)
-    logits = model.inference(image_node)
-    with tf.Session() as sess:
-        sess.run(tf.initialize_all_variables())
-        total_duration = 0
-        for epoch in range(1,FLAGS.epoch+1):
-            start_time = time.time()
-            for file_index in range(5):
-                print('Epoch %d: %s' % (epoch,filenames[file_index]))
-                reader = Cifar10Reader(filenames[file_index])
-                for index in range(10000):
-                    image = reader.read(index)
-                    logits_value = sess.run([logits],feed_dict={
-                        train_placeholder:image.byte_array,
-                    })
-                    if index % 1000 ==0:
-                        print('[%d]: %r'% (image.label,logits_value))
-            reader.close()
-            duration = time.time() - start_time
-            total_duration += duration
-            print('epoch %d duration = %d sec'%(epoch,duration))
-            tf.train.SummaryWriter(FLAGS.checkpoint_dir,sess.graph)
-        print('Total duration = %d sec'% total_duration)
-if __name__ == '__main__':
-    tf.app.run()
-```
-しかし、inference.py を実行すると
-```ここに言語を入力
-File "inference.py", line 44, in main
-    image = reader.read(index)
-  File "/Users/XXX/Desktop/cifar/reader.py", line 43, in read
-    self.bytestream.seek(record_bytes * index,0)
-AttributeError: 'Cifar10Reader' object has no attribute 'bytestream'
-```
 というエラーが出てしまいました。
 Cifar10Reader にbytestream を引数として持たせなければならない、という意味ですよね？でも参考urlの書き方ではCifar10Readerクラスの引数にbytestream を指定していません。

情報の追加

2017/04/11 11:42

投稿

退会済みユーザー

スコア0

test CHANGED Viewed

File without changes

test CHANGED Viewed

@@ -1,5 +1,7 @@
 AttributeError: 'Cifar10Reader' object has no attribute 'bytestream'  のエラー が出ました。
+reader.py
 ```ここに言語を入力
 # coding: utf-8
@@ -110,15 +112,387 @@
+他に使用しているスクリプトは
-しかし、
+png10.py
 ```ここに言語を入力
+# coding: utf-8
+from __future__ import absolute_import
+from __future__ import division
+from __future__ import print_function
+import os
+import numpy as np
+import tensorflow as tf
+from PIL import Image
+from reader import Cifar10Reader
+FLAGS = tf.app.flags.FLAGS
+tf.app.flags.DEFINE_string('file',None,"処理するファイルのパス")
+tf.app.flags.DEFINE_integer('offset',0,"読み飛ばすレコード数")
+tf.app.flags.DEFINE_integer('length',16,"読み込んで変換するレコード数")
+basename = os.path.basename(FLAGS.file)
+path = os.path.dirname(FLAGS.file)
+reader = Cifar10Reader(FLAGS.file)
+stop = FLAGS.offset + FLAGS.length
+for index in range(FLAGS.offset,stop):
+    image = reader.read(index)
+    print('label: %d' % image.label)
+    imageshow = Image.fromarray(image.byte_array.astype(np.unit8))
+    file_name = '%s-%02d-%d.png' % (basename,index,image.label)
+    file = os.path.join(path,file_name)
+    with open(file,mode='wb') as out:
+        imageshow.save(out,format='png')
+reader.close()
+```
+model.py
+```ここに言語を入力
+# coding: utf-8
+from __future__ import absolute_import
+from __future__ import division
+from __future__ import print_function
+import tensorflow as tf
+NUM_CLASSES = 10
+def _get_weights(shape,stddev=1.0):
+  var = tf.get_variable(
+      'weights',
+      shape,
+      initializer=tf.truncated_normal_initializer(stddev=stddev)
+  )
+  return var
+def _get_biases(shape,value=0.0):
+  var = tf.get_variable(
+      'biases',
+      shape,
+      initializer=tf.constant_initializer(value)
+  )
+  return var
+def inference(image_node):
+   #  conv1
+   with tf.variable_scope('conv1') as scope:
+     weights = _get_weights(shape=[5,5,3,64],stddev=1e-4)
+     conv = tf.nn.conv2d(image_node,weights,[1,1,1,1],padding='SAME')
+     biases = _get_biases([64],value=0.1)
+     bias = tf.nn.bias_add(conv,biases)
+     conv1 = tf.nn.relu(bias,name=scope.name)
+   #  pool
+   pool1 = tf.nn.max_pool(conv1,ksize=[1,3,3,1],strides=[1,2,2,1],padding='SAME',name='pool1')
+   #  conv2
+   with tf.variable_scope('conv2') as scope:
+     weights = _get_weights(shape=[5,5,64,64],stddev=1e-4)
+     conv = tf.nn.conv2d(pool1,weights,[1,1,1,1],padding='SAME')
+     biases = _get_biases([64],value=0.1)
+     bias = tf.nn.bias_add(conv,biases)
+     conv2 = tf.nn.relu(bias,name=scope.name)
+   #  pool2
+   pool2 = tf.nn.max_pool(conv2,ksize=[1,3,3,1],strides=[1,2,2,1],padding='SAME',name='pool2')
+   reshape = tf.reshape(pool2,[1,-1])
+   dim = reshape.get_shape()[1].value
+   #  fc3
+   with tf.variable_scope('fc3') as scope:
+       weights = _get_weights(shape=[dim,384],stddev=0.04)
+       biases = _get_biases([384],value=0.1)
+       fc3 = tf.nn.relu(
+           tf.matmul(reshape,weights) + biases,
+           name=scope.name
+       )
+   #  fc4
+   with tf.variable_scope('fc4') as scope:
+       weights = _get_weights(shape=[384,192],stddev=0.04)
+       biases = _get_biases([192],value=0.1)
+       fc4 = tf.nn.relu(tf.matmul(fc3,weights) + biases,name=scope.name)
+   #   output
+   with tf.variable_scope('output') as scope:
+       weights = _get_weights(shape=[192,NUM_CLASSES],stddev=1/192.0)
+       biases = _get_biases([NUM_CLASSES],value=0.0)
+       logits = tf.add(tf.matmul(fc4,weights),biases,name='logits')
+   return logits
+```
+inference.py
+```ここに言語を入力
+# coding: utf-8
+from __future__ import absolute_import
+from __future__ import division
+from __future__ import print_function
+import os
+import time
+import tensorflow as tf
+import model as model
+from reader import Cifar10Reader
+FLAGS = tf.app.flags.FLAGS
+tf.app.flags.DEFINE_integer('epoch',30,"訓練するEpoch数")
+tf.app.flags.DEFINE_string('data_dir','./data/',"訓練データのディレクトリ")
+tf.app.flags.DEFINE_string('checkpoint_dir','./checkpoints/',"チェックポイントを保存するディレクトリ")
+filenames = [
+    os.path.join(FLAGS.data_dir,'data_batch_%d.bin' % i) for i in range(1,6)
+]
+def main(argv=None):
+    train_placeholder = tf.placeholder(tf.float32,shape=[32,32,3],name='input_image')
+    image_node = tf.expand_dims(train_placeholder,0)
+    logits = model.inference(image_node)
+    with tf.Session() as sess:
+        sess.run(tf.initialize_all_variables())
+        total_duration = 0
+        for epoch in range(1,FLAGS.epoch+1):
+            start_time = time.time()
+            for file_index in range(5):
+                print('Epoch %d: %s' % (epoch,filenames[file_index]))
+                reader = Cifar10Reader(filenames[file_index])
+                for index in range(10000):
+                    image = reader.read(index)
+                    logits_value = sess.run([logits],feed_dict={
+                        train_placeholder:image.byte_array,
+                    })
+                    if index % 1000 ==0:
+                        print('[%d]: %r'% (image.label,logits_value))
+            reader.close()
+            duration = time.time() - start_time
+            total_duration += duration
+            print('epoch %d duration = %d sec'%(epoch,duration))
+            tf.train.SummaryWriter(FLAGS.checkpoint_dir,sess.graph)
+        print('Total duration = %d sec'% total_duration)
+if __name__ == '__main__':
+    tf.app.run()
+```
+しかし、inference.py を実行すると
+```ここに言語を入力
 File "inference.py", line 44, in main
     image = reader.read(index)
-  File "/Users/hasuikeyuri/Desktop/cifar/reader.py", line 43, in read
+  File "/Users/XXX/Desktop/cifar/reader.py", line 43, in read
     self.bytestream.seek(record_bytes * index,0)