diff --git a/paddlehub/datasets/canvas.py b/paddlehub/datasets/canvas.py
index e32376d5cd13e540e9b959cf394c6c49ee057c7a..22e1aefe520812a3d01b9b0cc47d94659af88fd1 100644
--- a/paddlehub/datasets/canvas.py
+++ b/paddlehub/datasets/canvas.py
@@ -19,8 +19,8 @@ from typing import Callable
 import paddle
 import numpy as np
 
+import paddlehub.env as hubenv
 from paddlehub.vision.utils import get_img_file
-from paddlehub.env import DATA_HOME
 from paddlehub.utils.download import download_data
 
 
@@ -47,7 +47,7 @@ class Canvas(paddle.io.Dataset):
         elif self.mode == 'test':
             self.file = 'test'
 
-        self.file = os.path.join(DATA_HOME, 'canvas', self.file)
+        self.file = os.path.join(hubenv.DATA_HOME, 'canvas', self.file)
         self.data = get_img_file(self.file)
 
     def __getitem__(self, idx: int) -> np.ndarray:
diff --git a/paddlehub/datasets/flowers.py b/paddlehub/datasets/flowers.py
index 5935f83dfb715e5648c7a7f4ca5725953fa40649..91452b9545002de1f755bc80836913a8a41d94f9 100644
--- a/paddlehub/datasets/flowers.py
+++ b/paddlehub/datasets/flowers.py
@@ -14,14 +14,18 @@
 # limitations under the License.
 
 import os
+from typing import Callable, Tuple
 
 import paddle
+import numpy as np
 
-from paddlehub.env import DATA_HOME
+import paddlehub.env as hubenv
+from paddlehub.utils.download import download_data
 
 
+@download_data(url='https://bj.bcebos.com/paddlehub-dataset/flower_photos.tar.gz')
 class Flowers(paddle.io.Dataset):
-    def __init__(self, transforms=None, mode='train'):
+    def __init__(self, transforms: Callable, mode: str = 'train'):
         self.mode = mode
         self.transforms = transforms
         self.num_classes = 5
@@ -32,14 +36,14 @@ class Flowers(paddle.io.Dataset):
             self.file = 'test_list.txt'
         else:
             self.file = 'validate_list.txt'
-        self.file = os.path.join(DATA_HOME, 'flower_photos', self.file)
+        self.file = os.path.join(hubenv.DATA_HOME, 'flower_photos', self.file)
 
         with open(self.file, 'r') as file:
             self.data = file.read().split('\n')
 
-    def __getitem__(self, idx):
+    def __getitem__(self, idx) -> Tuple[np.ndarray, int]:
         img_path, grt = self.data[idx].split(' ')
-        img_path = os.path.join(DATA_HOME, 'flower_photos', img_path)
+        img_path = os.path.join(hubenv.DATA_HOME, 'flower_photos', img_path)
         im = self.transforms(img_path)
         return im, int(grt)
 
diff --git a/paddlehub/datasets/minicoco.py b/paddlehub/datasets/minicoco.py
index 552996c7e3c76dffac4c6de4e9043a7596f8d5e4..01d3d00dc45c0cc237ef135a5434c1f99a54f4a2 100644
--- a/paddlehub/datasets/minicoco.py
+++ b/paddlehub/datasets/minicoco.py
@@ -19,8 +19,8 @@ from typing import Callable
 import paddle
 import numpy as np
 
+import paddlehub.env as hubenv
 from paddlehub.vision.utils import get_img_file
-from paddlehub.env import DATA_HOME
 from paddlehub.utils.download import download_data
 
 
@@ -46,8 +46,8 @@ class MiniCOCO(paddle.io.Dataset):
             self.file = 'train'
         elif self.mode == 'test':
             self.file = 'test'
-        self.file = os.path.join(DATA_HOME, 'minicoco', self.file)
-        self.style_file = os.path.join(DATA_HOME, 'minicoco', '21styles')
+        self.file = os.path.join(hubenv.DATA_HOME, 'minicoco', self.file)
+        self.style_file = os.path.join(hubenv.DATA_HOME, 'minicoco', '21styles')
         self.data = get_img_file(self.file)
         self.style = get_img_file(self.style_file)