fix global_step scope problem

examples/HED/hed.py

@@ -44,14 +44,11 @@ Usage:
 """
 
 class Model(ModelDesc):
-    def __init__(self, is_training=True):
-        self.isTrain = is_training
-
     def _get_input_vars(self):
         return [InputVar(tf.float32, [None, None, None] + [3], 'image'),
                 InputVar(tf.int32, [None, None, None], 'edgemap') ]
 
-    def _build_graph(self, input_vars, is_training):
+    def _build_graph(self, input_vars):
         image, edgemap = input_vars
         image = image - tf.constant([104, 116, 122], dtype='float32')
 
@@ -91,8 +88,12 @@ class Model(ModelDesc):
             l = Conv2D('conv5_3', l, 512)
             b5 = branch('branch5', l, 16)
 
-        final_map = tf.squeeze(tf.mul(0.2, b1 + b2 + b3 + b4 + b5),
-                [3], name='predmap')
+        final_map = Conv2D('convfcweight',
+                tf.concat(3, [b1, b2, b3, b4, b5]), 1, 1,
+                W_init=tf.constant_initializer(0.2), use_bias=False)
+        final_map = tf.squeeze(final_map, [3], name='predmap')
+        #final_map = tf.squeeze(tf.mul(0.2, b1 + b2 + b3 + b4 + b5),
+                #[3], name='predmap')
         costs = []
         for idx, b in enumerate([b1, b2, b3, b4, b5, final_map]):
             output = tf.nn.sigmoid(b, name='output{}'.format(idx+1))
@@ -114,6 +115,10 @@ class Model(ModelDesc):
         add_param_summary([('.*/W', ['histogram'])])   # monitor W
         self.cost = tf.add_n(costs, name='cost')
 
+    def get_gradient_processor(self):
+        return [ScaleGradient([('convfc.*', 0.1), ('conv5_.*', 100)]),
+                SummaryGradient()]
+
 def get_data(name):
     isTrain = name == 'train'
     ds = dataset.BSDS500(name, shuffle=True)
@@ -149,8 +154,8 @@ def get_data(name):
     ds = AugmentImageComponents(ds, shape_aug, (0, 1))
 
     def f(m):
-        m[m>=0.49] = 1
-        m[m<0.49] = 0
+        m[m>=0.51] = 1
+        m[m<0.51] = 0
         return m
     ds = MapDataComponent(ds, f, 1)
 
@@ -161,9 +166,10 @@ def get_data(name):
             imgaug.GaussianNoise(),
         ]
         ds = AugmentImageComponent(ds, augmentors)
-    ds = BatchDataByShape(ds, 8, idx=0)
-    if isTrain:
+        ds = BatchDataByShape(ds, 8, idx=0)
         ds = PrefetchDataZMQ(ds, 1)
+    else:
+        ds = BatchData(ds, 1)
     return ds
 
 def view_data():
@@ -180,27 +186,28 @@ def view_data():
 def get_config():
     logger.auto_set_dir()
     dataset_train = get_data('train')
-    step_per_epoch = dataset_train.size() * 100
+    step_per_epoch = dataset_train.size() * 40
     dataset_val = get_data('val')
     #dataset_test = get_data('test')
 
-    lr = tf.Variable(1e-5, trainable=False, name='learning_rate')
+    lr = tf.Variable(5e-6, trainable=False, name='learning_rate')
     tf.scalar_summary('learning_rate', lr)
 
     return TrainConfig(
         dataset=dataset_train,
-        optimizer=tf.train.AdamOptimizer(lr, epsilon=1e-3),
+        #optimizer=tf.train.AdamOptimizer(lr, epsilon=1e-3),
+        optimizer=tf.train.MomentumOptimizer(lr, 0.9),
         callbacks=Callbacks([
             StatPrinter(),
             ModelSaver(),
-            ScheduledHyperParamSetter('learning_rate', [(25, 3e-6)]),
+            ScheduledHyperParamSetter('learning_rate', [(100, 3e-6), (200, 8e-7)]),
             HumanHyperParamSetter('learning_rate'),
             InferenceRunner(dataset_val,
                             BinaryClassificationStats('prediction', 'edgemap'))
         ]),
         model=Model(),
         step_per_epoch=step_per_epoch,
-        max_epoch=100,
+        max_epoch=300,
     )
 
 def run(model_path, image_path):

tensorpack/dataflow/imgaug/noname.py

@@ -60,7 +60,8 @@ class Resize(ImageAugmentor):
 
 class RandomResize(ImageAugmentor):
     """ randomly rescale w and h of the image"""
-    def __init__(self, xrange, yrange, minimum=(0,0), aspect_ratio_thres=0.15):
+    def __init__(self, xrange, yrange, minimum=(0,0), aspect_ratio_thres=0.15,
+            interp=cv2.INTER_CUBIC):
         """
         :param xrange: (min, max) scaling ratio
         :param yrange: (min, max) scaling ratio
@@ -88,5 +89,5 @@ class RandomResize(ImageAugmentor):
                 return img.shape[1], img.shape[0]
 
     def _augment(self, img, dsize):
-        return cv2.resize(img, dsize, interpolation=cv2.INTER_CUBIC)
+        return cv2.resize(img, dsize, interpolation=self.interp)
 

tensorpack/tfutils/common.py

@@ -45,8 +45,8 @@ def get_global_step_var():
         scope = tf.get_variable_scope()
         assert scope.name == '', \
                 "Creating global_step_var under a variable scope would cause problems!"
-        var = tf.Variable(
-            0, trainable=False, name=GLOBAL_STEP_OP_NAME)
+        var = tf.get_variable(GLOBAL_STEP_OP_NAME, shape=[],
+                initializer=tf.constant_initializer(), trainable=False)
         return var
 
 def get_global_step():