diff --git a/fluid/object_detection/mobilenet_ssd.py b/fluid/object_detection/mobilenet_ssd.py
index c9a8e7793fc4db347243a6501544844bc37dff36..21869647aa261a1baacbe934453c8af4416b75b7 100644
--- a/fluid/object_detection/mobilenet_ssd.py
+++ b/fluid/object_detection/mobilenet_ssd.py
@@ -13,7 +13,7 @@ def conv_bn(input,
             num_groups=1,
             act='relu',
             use_cudnn=True):
-    parameter_attr = ParamAttr(initializer=MSRA())
+    parameter_attr = ParamAttr(learning_rate=0.1, initializer=MSRA())
     conv = fluid.layers.conv2d(
         input=input,
         num_filters=num_filters,
@@ -25,7 +25,14 @@ def conv_bn(input,
         use_cudnn=use_cudnn,
         param_attr=parameter_attr,
         bias_attr=False)
-    return fluid.layers.batch_norm(input=conv, act=act, epsilon=0.00001)
+    parameter_attr = ParamAttr(learning_rate=0.1, initializer=MSRA())
+    bias_attr = ParamAttr(learning_rate=0.2)
+    return fluid.layers.batch_norm(
+        input=conv,
+        act=act,
+        epsilon=0.00001,
+        param_attr=parameter_attr,
+        bias_attr=bias_attr)
 
 
 def depthwise_separable(input, num_filters1, num_filters2, num_groups, stride,
diff --git a/fluid/object_detection/reader.py b/fluid/object_detection/reader.py
index 29f218941029eee34acf4492911b58902f7f0c9c..4e680c29997b432c14b92ea641aa9f956de41292 100644
--- a/fluid/object_detection/reader.py
+++ b/fluid/object_detection/reader.py
@@ -150,6 +150,7 @@ def _reader_creator(settings, file_list, mode, shuffle):
                     img = np.swapaxes(img, 1, 2)
                     img = np.swapaxes(img, 1, 0)
 
+                img = img[[2, 1, 0], :, :]
                 img = img.astype('float32')
                 img -= settings.img_mean
                 img = img.flatten()
diff --git a/fluid/object_detection/train.py b/fluid/object_detection/train.py
index 0ad7f59e85f475b395a97812d93f4c639ca00bc3..9d2dbadccf4dd33db1dee2b71fcdcfa63f4246e8 100644
--- a/fluid/object_detection/train.py
+++ b/fluid/object_detection/train.py
@@ -45,13 +45,14 @@ def train(train_file_list,
             evaluate_difficult=False,
             ap_version='11point')
 
-    optimizer = fluid.optimizer.DecayedAdagrad(
+    optimizer = fluid.optimizer.Momentum(
         learning_rate=fluid.layers.exponential_decay(
             learning_rate=learning_rate,
             decay_steps=40000,
             decay_rate=0.1,
             staircase=True),
-        regularization=fluid.regularizer.L2Decay(0.0005), )
+        momentum=0.9,
+        regularization=fluid.regularizer.L2Decay(0.00005), )
 
     optimizer.minimize(loss)
 
@@ -59,7 +60,7 @@ def train(train_file_list,
     exe = fluid.Executor(place)
     exe.run(fluid.default_startup_program())
 
-    load_model.load_and_set_vars(place)
+    load_model.load_paddlev1_vars(place)
     train_reader = paddle.batch(
         reader.train(data_args, train_file_list), batch_size=batch_size)
     test_reader = paddle.batch(