diff --git a/examples/ocr/data.py b/examples/ocr/data.py
index 00c4b1ac5a1116443ce5a8ecaeeb35e66782e521..cb0e13f608af5abb58ae2f5244e4d089ce663f33 100644
--- a/examples/ocr/data.py
+++ b/examples/ocr/data.py
@@ -29,40 +29,6 @@ TRAIN_LIST_FILE_NAME = "train.list"
 TEST_LIST_FILE_NAME = "test.list"
 
 
-class BatchCompose(object):
-    def __init__(self, transforms=[]):
-        self.transforms = transforms
-
-    def __call__(self, data):
-        for f in self.transforms:
-            try:
-                data = f(data)
-            except Exception as e:
-                stack_info = traceback.format_exc()
-                logger.info("fail to perform batch transform [{}] with error: "
-                            "{} and stack:\n{}".format(f, e, str(stack_info)))
-                raise e
-        # sample list to batch data
-        batch = list(zip(*data))
-        return batch
-
-
-class Compose(object):
-    def __init__(self, transforms=[]):
-        self.transforms = transforms
-
-    def __call__(self, *data):
-        for f in self.transforms:
-            try:
-                data = f(*data)
-            except Exception as e:
-                stack_info = traceback.format_exc()
-                logger.info("fail to perform transform [{}] with error: "
-                            "{} and stack:\n{}".format(f, e, str(stack_info)))
-                raise e
-        return data
-
-
 class Resize(object):
     def __init__(self, height=48):
         self.interp = Image.NEAREST  # Image.ANTIALIAS
diff --git a/examples/ocr/eval.py b/examples/ocr/eval.py
index e3c487da64baf41c68904263e840701ac91c8545..2fd22751694f9def9988bb24d9292ad82a442337 100644
--- a/examples/ocr/eval.py
+++ b/examples/ocr/eval.py
@@ -20,6 +20,7 @@ import paddle.fluid.profiler as profiler
 import paddle.fluid as fluid
 
 from hapi.model import Input, set_device
+from hapi.vision.transforms import BatchCompose
 
 from utility import add_arguments, print_arguments
 from utility import SeqAccuracy, MyProgBarLogger, SeqBeamAccuracy
@@ -73,7 +74,7 @@ def main(FLAGS):
     model.load(FLAGS.init_model)
 
     test_dataset = data.test()
-    test_collate_fn = data.BatchCompose(
+    test_collate_fn = BatchCompose(
         [data.Resize(), data.Normalize(), data.PadTarget()])
     test_sampler = data.MyBatchSampler(
         test_dataset,
@@ -122,7 +123,7 @@ def beam_search(FLAGS):
     model.load(FLAGS.init_model)
 
     test_dataset = data.test()
-    test_collate_fn = data.BatchCompose(
+    test_collate_fn = BatchCompose(
         [data.Resize(), data.Normalize(), data.PadTarget()])
     test_sampler = data.MyBatchSampler(
         test_dataset,
diff --git a/examples/ocr/predict.py b/examples/ocr/predict.py
index d1a66f7b053fcd83bb548eddf7bbcfcf5ca74c8e..df0f21e9ea16e1442159ed049d4867c066d4d4e4 100644
--- a/examples/ocr/predict.py
+++ b/examples/ocr/predict.py
@@ -27,6 +27,7 @@ import paddle.fluid as fluid
 
 from hapi.model import Input, set_device
 from hapi.datasets.folder import ImageFolder
+from hapi.vision.transforms import BatchCompose
 
 from utility import add_arguments, print_arguments
 from utility import postprocess, index2word
@@ -67,7 +68,7 @@ def main(FLAGS):
 
     fn = lambda p: Image.open(p).convert('L')
     test_dataset = ImageFolder(FLAGS.image_path, loader=fn)
-    test_collate_fn = data.BatchCompose([data.Resize(), data.Normalize()])
+    test_collate_fn = BatchCompose([data.Resize(), data.Normalize()])
     test_loader = fluid.io.DataLoader(
         test_dataset,
         places=device,
diff --git a/examples/ocr/train.py b/examples/ocr/train.py
index 789edcafe5859bf031069d3e23697cd1e56b4399..423e27d6d2171e125d3108257634997393caa0fc 100644
--- a/examples/ocr/train.py
+++ b/examples/ocr/train.py
@@ -25,6 +25,7 @@ import paddle.fluid.profiler as profiler
 import paddle.fluid as fluid
 
 from hapi.model import Input, set_device
+from hapi.vision.transforms import BatchCompose
 
 from utility import add_arguments, print_arguments
 from utility import SeqAccuracy, MyProgBarLogger
@@ -97,7 +98,7 @@ def main(FLAGS):
         labels=labels)
 
     train_dataset = data.train()
-    train_collate_fn = data.BatchCompose(
+    train_collate_fn = BatchCompose(
         [data.Resize(), data.Normalize(), data.PadTarget()])
     train_sampler = data.MyBatchSampler(
         train_dataset, batch_size=FLAGS.batch_size, shuffle=True)
@@ -109,7 +110,7 @@ def main(FLAGS):
         return_list=True,
         collate_fn=train_collate_fn)
     test_dataset = data.test()
-    test_collate_fn = data.BatchCompose(
+    test_collate_fn = BatchCompose(
         [data.Resize(), data.Normalize(), data.PadTarget()])
     test_sampler = data.MyBatchSampler(
         test_dataset,