Make vision datasets return PIL.Image as default (#28264)

* return pil image as default according backend

python/paddle/hapi/callbacks.py

@@ -296,12 +296,17 @@ class ProgBarLogger(Callback):
         .. code-block:: python
 
             import paddle
+            import paddle.vision.transforms as T
             from paddle.static import InputSpec
 
             inputs = [InputSpec([-1, 1, 28, 28], 'float32', 'image')]
             labels = [InputSpec([None, 1], 'int64', 'label')]
 
-            train_dataset = paddle.vision.datasets.MNIST(mode='train')
+            transform = T.Compose([
+                T.Transpose(),
+                T.Normalize([127.5], [127.5])
+            ])
+            train_dataset = paddle.vision.datasets.MNIST(mode='train', transform=transform)
 
             lenet = paddle.vision.LeNet()
             model = paddle.Model(lenet,
@@ -432,12 +437,17 @@ class ModelCheckpoint(Callback):
         .. code-block:: python
 
             import paddle
+            import paddle.vision.transforms as T
             from paddle.static import InputSpec
 
             inputs = [InputSpec([-1, 1, 28, 28], 'float32', 'image')]
             labels = [InputSpec([None, 1], 'int64', 'label')]
 
-            train_dataset = paddle.vision.datasets.MNIST(mode='train')
+            transform = T.Compose([
+                T.Transpose(),
+                T.Normalize([127.5], [127.5])
+            ])
+            train_dataset = paddle.vision.datasets.MNIST(mode='train', transform=transform)
 
             lenet = paddle.vision.LeNet()
             model = paddle.Model(lenet,
@@ -484,13 +494,18 @@ class VisualDL(Callback):
         .. code-block:: python
 
             import paddle
+            import paddle.vision.transforms as T
             from paddle.static import InputSpec
 
             inputs = [InputSpec([-1, 1, 28, 28], 'float32', 'image')]
             labels = [InputSpec([None, 1], 'int64', 'label')]
 
-            train_dataset = paddle.vision.datasets.MNIST(mode='train')
-            eval_dataset = paddle.vision.datasets.MNIST(mode='test')
+            transform = T.Compose([
+                T.Transpose(),
+                T.Normalize([127.5], [127.5])
+            ])
+            train_dataset = paddle.vision.datasets.MNIST(mode='train', transform=transform)
+            eval_dataset = paddle.vision.datasets.MNIST(mode='test', transform=transform)
 
             net = paddle.vision.LeNet()
             model = paddle.Model(net, inputs, labels)

python/paddle/hapi/model.py

@@ -837,6 +837,7 @@ class Model(object):
 
         import paddle
         import paddle.nn as nn
+        import paddle.vision.transforms as T
         from paddle.static import InputSpec
 
         device = paddle.set_device('cpu') # or 'gpu'
@@ -858,7 +859,11 @@ class Model(object):
                       paddle.nn.CrossEntropyLoss(),
                       paddle.metric.Accuracy())
         
-        data = paddle.vision.datasets.MNIST(mode='train')
+        transform = T.Compose([
+            T.Transpose(),
+            T.Normalize([127.5], [127.5])
+        ])
+        data = paddle.vision.datasets.MNIST(mode='train', transform=transform)
         model.fit(data, epochs=2, batch_size=32, verbose=1)
     """
 
@@ -1067,6 +1072,7 @@ class Model(object):
 
                 import paddle
                 import paddle.nn as nn
+                import paddle.vision.transforms as T
                 from paddle.static import InputSpec
 
                 class Mnist(nn.Layer):
@@ -1093,7 +1099,13 @@ class Model(object):
                 optim = paddle.optimizer.SGD(learning_rate=1e-3,
                     parameters=model.parameters())
                 model.prepare(optim, paddle.nn.CrossEntropyLoss())
-                data = paddle.vision.datasets.MNIST(mode='train')
+                
+                transform = T.Compose([
+                    T.Transpose(),
+                    T.Normalize([127.5], [127.5])
+                ])
+                data = paddle.vision.datasets.MNIST(mode='train', transform=transform)
+                
                 model.fit(data, epochs=1, batch_size=32, verbose=0)
                 model.save('checkpoint/test')  # save for training
                 model.save('inference_model', False)  # save for inference
@@ -1353,14 +1365,19 @@ class Model(object):
             .. code-block:: python
 
               import paddle
+              import paddle.vision.transforms as T
               from paddle.static import InputSpec
 
               dynamic = True
               device = paddle.set_device('cpu') # or 'gpu'
               paddle.disable_static(device) if dynamic else None
               
-              train_dataset = paddle.vision.datasets.MNIST(mode='train')
-              val_dataset = paddle.vision.datasets.MNIST(mode='test')
+              transform = T.Compose([
+                  T.Transpose(),
+                  T.Normalize([127.5], [127.5])
+              ])
+              train_dataset = paddle.vision.datasets.MNIST(mode='train', transform=transform)
+              val_dataset = paddle.vision.datasets.MNIST(mode='test', transform=transform)
            
               input = InputSpec([None, 1, 28, 28], 'float32', 'image')
               label = InputSpec([None, 1], 'int64', 'label')
@@ -1386,16 +1403,21 @@ class Model(object):
             .. code-block:: python
 
               import paddle
+              import paddle.vision.transforms as T
               from paddle.static import InputSpec
 
               dynamic = True
               device = paddle.set_device('cpu') # or 'gpu'
               paddle.disable_static(device) if dynamic else None
               
-              train_dataset = paddle.vision.datasets.MNIST(mode='train')
+              transform = T.Compose([
+                    T.Transpose(),
+                    T.Normalize([127.5], [127.5])
+                ])
+              train_dataset = paddle.vision.datasets.MNIST(mode='train', transform=transform)
               train_loader = paddle.io.DataLoader(train_dataset,
                   places=device, batch_size=64)
-              val_dataset = paddle.vision.datasets.MNIST(mode='test')
+              val_dataset = paddle.vision.datasets.MNIST(mode='test', transform=transform)
               val_loader = paddle.io.DataLoader(val_dataset,
                   places=device, batch_size=64)
            
@@ -1522,10 +1544,15 @@ class Model(object):
         .. code-block:: python
 
             import paddle
+            import paddle.vision.transforms as T
             from paddle.static import InputSpec
 
             # declarative mode
-            val_dataset = paddle.vision.datasets.MNIST(mode='test')
+            transform = T.Compose([
+                    T.Transpose(),
+                    T.Normalize([127.5], [127.5])
+                ])
+            val_dataset = paddle.vision.datasets.MNIST(mode='test', transform=transform)
 
             input = InputSpec([-1, 1, 28, 28], 'float32', 'image')
             label = InputSpec([None, 1], 'int64', 'label')

python/paddle/tests/test_callbacks.py

@@ -24,6 +24,7 @@ from paddle import Model
 from paddle.static import InputSpec
 from paddle.vision.models import LeNet
 from paddle.hapi.callbacks import config_callbacks
+import paddle.vision.transforms as T
 
 
 class TestCallbacks(unittest.TestCase):
@@ -112,8 +113,11 @@ class TestCallbacks(unittest.TestCase):
         inputs = [InputSpec([-1, 1, 28, 28], 'float32', 'image')]
         labels = [InputSpec([None, 1], 'int64', 'label')]
 
-        train_dataset = paddle.vision.datasets.MNIST(mode='train')
-        eval_dataset = paddle.vision.datasets.MNIST(mode='test')
+        transform = T.Compose([T.Transpose(), T.Normalize([127.5], [127.5])])
+        train_dataset = paddle.vision.datasets.MNIST(
+            mode='train', transform=transform)
+        eval_dataset = paddle.vision.datasets.MNIST(
+            mode='test', transform=transform)
 
         net = paddle.vision.LeNet()
         model = paddle.Model(net, inputs, labels)

python/paddle/tests/test_dataset_cifar.py

@@ -27,10 +27,11 @@ class TestCifar10Train(unittest.TestCase):
         # long time, randomly check 1 sample
         idx = np.random.randint(0, 50000)
         data, label = cifar[idx]
+        data = np.array(data)
         self.assertTrue(len(data.shape) == 3)
-        self.assertTrue(data.shape[0] == 3)
+        self.assertTrue(data.shape[2] == 3)
         self.assertTrue(data.shape[1] == 32)
-        self.assertTrue(data.shape[2] == 32)
+        self.assertTrue(data.shape[0] == 32)
         self.assertTrue(0 <= int(label) <= 9)
 
 
@@ -43,12 +44,30 @@ class TestCifar10Test(unittest.TestCase):
         # long time, randomly check 1 sample
         idx = np.random.randint(0, 10000)
         data, label = cifar[idx]
+        data = np.array(data)
         self.assertTrue(len(data.shape) == 3)
-        self.assertTrue(data.shape[0] == 3)
+        self.assertTrue(data.shape[2] == 3)
         self.assertTrue(data.shape[1] == 32)
-        self.assertTrue(data.shape[2] == 32)
+        self.assertTrue(data.shape[0] == 32)
         self.assertTrue(0 <= int(label) <= 9)
 
+        # test cv2 backend
+        cifar = Cifar10(mode='test', backend='cv2')
+        self.assertTrue(len(cifar) == 10000)
+
+        # traversal whole dataset may cost a
+        # long time, randomly check 1 sample
+        idx = np.random.randint(0, 10000)
+        data, label = cifar[idx]
+        self.assertTrue(len(data.shape) == 3)
+        self.assertTrue(data.shape[2] == 3)
+        self.assertTrue(data.shape[1] == 32)
+        self.assertTrue(data.shape[0] == 32)
+        self.assertTrue(0 <= int(label) <= 99)
+
+        with self.assertRaises(ValueError):
+            cifar = Cifar10(mode='test', backend=1)
+
 
 class TestCifar100Train(unittest.TestCase):
     def test_main(self):
@@ -59,10 +78,11 @@ class TestCifar100Train(unittest.TestCase):
         # long time, randomly check 1 sample
         idx = np.random.randint(0, 50000)
         data, label = cifar[idx]
+        data = np.array(data)
         self.assertTrue(len(data.shape) == 3)
-        self.assertTrue(data.shape[0] == 3)
+        self.assertTrue(data.shape[2] == 3)
         self.assertTrue(data.shape[1] == 32)
-        self.assertTrue(data.shape[2] == 32)
+        self.assertTrue(data.shape[0] == 32)
         self.assertTrue(0 <= int(label) <= 99)
 
 
@@ -75,12 +95,30 @@ class TestCifar100Test(unittest.TestCase):
         # long time, randomly check 1 sample
         idx = np.random.randint(0, 10000)
         data, label = cifar[idx]
+        data = np.array(data)
         self.assertTrue(len(data.shape) == 3)
-        self.assertTrue(data.shape[0] == 3)
+        self.assertTrue(data.shape[2] == 3)
         self.assertTrue(data.shape[1] == 32)
-        self.assertTrue(data.shape[2] == 32)
+        self.assertTrue(data.shape[0] == 32)
         self.assertTrue(0 <= int(label) <= 99)
 
+        # test cv2 backend
+        cifar = Cifar100(mode='test', backend='cv2')
+        self.assertTrue(len(cifar) == 10000)
+
+        # traversal whole dataset may cost a
+        # long time, randomly check 1 sample
+        idx = np.random.randint(0, 10000)
+        data, label = cifar[idx]
+        self.assertTrue(len(data.shape) == 3)
+        self.assertTrue(data.shape[2] == 3)
+        self.assertTrue(data.shape[1] == 32)
+        self.assertTrue(data.shape[0] == 32)
+        self.assertTrue(0 <= int(label) <= 99)
+
+        with self.assertRaises(ValueError):
+            cifar = Cifar100(mode='test', backend=1)
+
 
 if __name__ == '__main__':
     unittest.main()

python/paddle/tests/test_dataset_voc.py

@@ -32,6 +32,9 @@ class TestVOC2012Train(unittest.TestCase):
         # long time, randomly check 1 sample
         idx = np.random.randint(0, 3)
         image, label = voc2012[idx]
+        image = np.array(image)
+        label = np.array(label)
+
         self.assertTrue(len(image.shape) == 3)
         self.assertTrue(len(label.shape) == 2)
 
@@ -45,6 +48,9 @@ class TestVOC2012Valid(unittest.TestCase):
         # long time, randomly check 1 sample
         idx = np.random.randint(0, 1)
         image, label = voc2012[idx]
+        image = np.array(image)
+        label = np.array(label)
+
         self.assertTrue(len(image.shape) == 3)
         self.assertTrue(len(label.shape) == 2)
 
@@ -58,9 +64,27 @@ class TestVOC2012Test(unittest.TestCase):
         # long time, randomly check 1 sample
         idx = np.random.randint(0, 1)
         image, label = voc2012[idx]
+        image = np.array(image)
+        label = np.array(label)
+
+        self.assertTrue(len(image.shape) == 3)
+        self.assertTrue(len(label.shape) == 2)
+
+        # test cv2 backend
+        voc2012 = VOC2012(mode='test', backend='cv2')
+        self.assertTrue(len(voc2012) == 2)
+
+        # traversal whole dataset may cost a
+        # long time, randomly check 1 sample
+        idx = np.random.randint(0, 1)
+        image, label = voc2012[idx]
+
         self.assertTrue(len(image.shape) == 3)
         self.assertTrue(len(label.shape) == 2)
 
+        with self.assertRaises(ValueError):
+            voc2012 = VOC2012(mode='test', backend=1)
+
 
 if __name__ == '__main__':
     unittest.main()

python/paddle/tests/test_datasets.py

@@ -19,6 +19,7 @@ import tempfile
 import shutil
 import cv2
 
+import paddle.vision.transforms as T
 from paddle.vision.datasets import *
 from paddle.dataset.common import _check_exists_and_download
 
@@ -89,7 +90,8 @@ class TestFolderDatasets(unittest.TestCase):
 
 class TestMNISTTest(unittest.TestCase):
     def test_main(self):
-        mnist = MNIST(mode='test')
+        transform = T.Transpose()
+        mnist = MNIST(mode='test', transform=transform)
         self.assertTrue(len(mnist) == 10000)
 
         for i in range(len(mnist)):
@@ -103,7 +105,8 @@ class TestMNISTTest(unittest.TestCase):
 
 class TestMNISTTrain(unittest.TestCase):
     def test_main(self):
-        mnist = MNIST(mode='train')
+        transform = T.Transpose()
+        mnist = MNIST(mode='train', transform=transform)
         self.assertTrue(len(mnist) == 60000)
 
         for i in range(len(mnist)):
@@ -114,6 +117,22 @@ class TestMNISTTrain(unittest.TestCase):
             self.assertTrue(label.shape[0] == 1)
             self.assertTrue(0 <= int(label) <= 9)
 
+        # test cv2 backend
+        mnist = MNIST(mode='train', transform=transform, backend='cv2')
+        self.assertTrue(len(mnist) == 60000)
+
+        for i in range(len(mnist)):
+            image, label = mnist[i]
+            self.assertTrue(image.shape[0] == 1)
+            self.assertTrue(image.shape[1] == 28)
+            self.assertTrue(image.shape[2] == 28)
+            self.assertTrue(label.shape[0] == 1)
+            self.assertTrue(0 <= int(label) <= 9)
+            break
+
+        with self.assertRaises(ValueError):
+            mnist = MNIST(mode='train', transform=transform, backend=1)
+
 
 class TestFlowersTrain(unittest.TestCase):
     def test_main(self):
@@ -124,6 +143,7 @@ class TestFlowersTrain(unittest.TestCase):
         # long time, randomly check 1 sample
         idx = np.random.randint(0, 6149)
         image, label = flowers[idx]
+        image = np.array(image)
         self.assertTrue(len(image.shape) == 3)
         self.assertTrue(image.shape[2] == 3)
         self.assertTrue(label.shape[0] == 1)
@@ -138,6 +158,7 @@ class TestFlowersValid(unittest.TestCase):
         # long time, randomly check 1 sample
         idx = np.random.randint(0, 1020)
         image, label = flowers[idx]
+        image = np.array(image)
         self.assertTrue(len(image.shape) == 3)
         self.assertTrue(image.shape[2] == 3)
         self.assertTrue(label.shape[0] == 1)
@@ -152,10 +173,27 @@ class TestFlowersTest(unittest.TestCase):
         # long time, randomly check 1 sample
         idx = np.random.randint(0, 1020)
         image, label = flowers[idx]
+        image = np.array(image)
         self.assertTrue(len(image.shape) == 3)
         self.assertTrue(image.shape[2] == 3)
         self.assertTrue(label.shape[0] == 1)
 
+        # test cv2 backend
+        flowers = Flowers(mode='test', backend='cv2')
+        self.assertTrue(len(flowers) == 1020)
+
+        # traversal whole dataset may cost a
+        # long time, randomly check 1 sample
+        idx = np.random.randint(0, 1020)
+        image, label = flowers[idx]
+
+        self.assertTrue(len(image.shape) == 3)
+        self.assertTrue(image.shape[2] == 3)
+        self.assertTrue(label.shape[0] == 1)
+
+        with self.assertRaises(ValueError):
+            flowers = Flowers(mode='test', backend=1)
+
 
 if __name__ == '__main__':
     unittest.main()

python/paddle/vision/datasets/cifar.py

@@ -17,6 +17,7 @@ from __future__ import print_function
 import tarfile
 import numpy as np
 import six
+from PIL import Image
 from six.moves import cPickle as pickle
 
 import paddle
@@ -51,6 +52,10 @@ class Cifar10(Dataset):
         transform(callable): transform to perform on image, None for on transform.
         download(bool): whether to download dataset automatically if
             :attr:`data_file` is not set. Default True
+        backend(str, optional): Specifies which type of image to be returned: 
+            PIL.Image or numpy.ndarray. Should be one of {'pil', 'cv2'}. 
+            If this option is not set, will get backend from ``paddle.vsion.get_image_backend`` ,
+            default backend is 'pil'. Default: None.
 
     Returns:
         Dataset: instance of cifar-10 dataset
@@ -72,13 +77,14 @@ class Cifar10(Dataset):
                         nn.Softmax())
 
                 def forward(self, image, label):
-                    image = paddle.reshape(image, (3, -1))
+                    image = paddle.reshape(image, (1, -1))
                     return self.fc(image), label
 
             paddle.disable_static()
 
             normalize = Normalize(mean=[0.5, 0.5, 0.5],
-                                std=[0.5, 0.5, 0.5])
+                                  std=[0.5, 0.5, 0.5],
+                                  data_format='HWC')
             cifar10 = Cifar10(mode='train', transform=normalize)
 
             for i in range(10):
@@ -96,11 +102,20 @@ class Cifar10(Dataset):
                  data_file=None,
                  mode='train',
                  transform=None,
-                 download=True):
+                 download=True,
+                 backend=None):
         assert mode.lower() in ['train', 'test', 'train', 'test'], \
             "mode should be 'train10', 'test10', 'train100' or 'test100', but got {}".format(mode)
         self.mode = mode.lower()
 
+        if backend is None:
+            backend = paddle.vision.get_image_backend()
+        if backend not in ['pil', 'cv2']:
+            raise ValueError(
+                "Expected backend are one of ['pil', 'cv2'], but got {}"
+                .format(backend))
+        self.backend = backend
+
         self._init_url_md5_flag()
 
         self.data_file = data_file
@@ -143,8 +158,16 @@ class Cifar10(Dataset):
     def __getitem__(self, idx):
         image, label = self.data[idx]
         image = np.reshape(image, [3, 32, 32])
+        image = image.transpose([1, 2, 0])
+
+        if self.backend == 'pil':
+            image = Image.fromarray(image)
         if self.transform is not None:
             image = self.transform(image)
+
+        if self.backend == 'pil':
+            return image, np.array(label).astype('int64')
+
         return image.astype(self.dtype), np.array(label).astype('int64')
 
     def __len__(self):
@@ -163,6 +186,10 @@ class Cifar100(Cifar10):
         transform(callable): transform to perform on image, None for on transform.
         download(bool): whether to download dataset automatically if
             :attr:`data_file` is not set. Default True
+        backend(str, optional): Specifies which type of image to be returned: 
+            PIL.Image or numpy.ndarray. Should be one of {'pil', 'cv2'}. 
+            If this option is not set, will get backend from ``paddle.vsion.get_image_backend`` ,
+            default backend is 'pil'. Default: None.
 
     Returns:
         Dataset: instance of cifar-100 dataset
@@ -184,13 +211,14 @@ class Cifar100(Cifar10):
                         nn.Softmax())
 
                 def forward(self, image, label):
-                    image = paddle.reshape(image, (3, -1))
+                    image = paddle.reshape(image, (1, -1))
                     return self.fc(image), label
 
             paddle.disable_static()
 
             normalize = Normalize(mean=[0.5, 0.5, 0.5],
-                                std=[0.5, 0.5, 0.5])
+                                  std=[0.5, 0.5, 0.5],
+                                  data_format='HWC')
             cifar100 = Cifar100(mode='train', transform=normalize)
 
             for i in range(10):
@@ -208,8 +236,10 @@ class Cifar100(Cifar10):
                  data_file=None,
                  mode='train',
                  transform=None,
-                 download=True):
-        super(Cifar100, self).__init__(data_file, mode, transform, download)
+                 download=True,
+                 backend=None):
+        super(Cifar100, self).__init__(data_file, mode, transform, download,
+                                       backend)
 
     def _init_url_md5_flag(self):
         self.data_url = CIFAR100_URL

python/paddle/vision/datasets/flowers.py

@@ -56,6 +56,10 @@ class Flowers(Dataset):
         transform(callable): transform to perform on image, None for on transform.
         download(bool): whether to download dataset automatically if
             :attr:`data_file` is not set. Default True
+        backend(str, optional): Specifies which type of image to be returned: 
+            PIL.Image or numpy.ndarray. Should be one of {'pil', 'cv2'}. 
+            If this option is not set, will get backend from ``paddle.vsion.get_image_backend`` ,
+            default backend is 'pil'. Default: None.
 
     Examples:
         
@@ -67,7 +71,7 @@ class Flowers(Dataset):
 
             for i in range(len(flowers)):
                 sample = flowers[i]
-                print(sample[0].shape, sample[1])
+                print(sample[0].size, sample[1])
 
     """
 
@@ -77,9 +81,19 @@ class Flowers(Dataset):
                  setid_file=None,
                  mode='train',
                  transform=None,
-                 download=True):
+                 download=True,
+                 backend=None):
         assert mode.lower() in ['train', 'valid', 'test'], \
                 "mode should be 'train', 'valid' or 'test', but got {}".format(mode)
+
+        if backend is None:
+            backend = paddle.vision.get_image_backend()
+        if backend not in ['pil', 'cv2']:
+            raise ValueError(
+                "Expected backend are one of ['pil', 'cv2'], but got {}"
+                .format(backend))
+        self.backend = backend
+
         self.flag = MODE_FLAG_MAP[mode.lower()]
 
         self.data_file = data_file
@@ -122,11 +136,18 @@ class Flowers(Dataset):
         img_name = "jpg/image_%05d.jpg" % index
         img_ele = self.name2mem[img_name]
         image = self.data_tar.extractfile(img_ele).read()
+
+        if self.backend == 'pil':
+            image = Image.open(io.BytesIO(image))
+        elif self.backend == 'cv2':
             image = np.array(Image.open(io.BytesIO(image)))
 
         if self.transform is not None:
             image = self.transform(image)
 
+        if self.backend == 'pil':
+            return image, label.astype('int64')
+
         return image.astype(self.dtype), label.astype('int64')
 
     def __len__(self):

python/paddle/vision/datasets/mnist.py

@@ -18,6 +18,7 @@ import os
 import gzip
 import struct
 import numpy as np
+from PIL import Image
 
 import paddle
 from paddle.io import Dataset
@@ -48,6 +49,10 @@ class MNIST(Dataset):
         mode(str): 'train' or 'test' mode. Default 'train'.
         download(bool): whether to download dataset automatically if
             :attr:`image_path` :attr:`label_path` is not set. Default True
+        backend(str, optional): Specifies which type of image to be returned: 
+            PIL.Image or numpy.ndarray. Should be one of {'pil', 'cv2'}. 
+            If this option is not set, will get backend from ``paddle.vsion.get_image_backend`` ,
+            default backend is 'pil'. Default: None.
             
     Returns:
         Dataset: MNIST Dataset.
@@ -62,7 +67,7 @@ class MNIST(Dataset):
 
             for i in range(len(mnist)):
                 sample = mnist[i]
-                print(sample[0].shape, sample[1])
+                print(sample[0].size, sample[1])
 
     """
 
@@ -71,9 +76,19 @@ class MNIST(Dataset):
                  label_path=None,
                  mode='train',
                  transform=None,
-                 download=True):
+                 download=True,
+                 backend=None):
         assert mode.lower() in ['train', 'test'], \
                 "mode should be 'train' or 'test', but got {}".format(mode)
+
+        if backend is None:
+            backend = paddle.vision.get_image_backend()
+        if backend not in ['pil', 'cv2']:
+            raise ValueError(
+                "Expected backend are one of ['pil', 'cv2'], but got {}"
+                .format(backend))
+        self.backend = backend
+
         self.mode = mode.lower()
         self.image_path = image_path
         if self.image_path is None:
@@ -145,9 +160,17 @@ class MNIST(Dataset):
 
     def __getitem__(self, idx):
         image, label = self.images[idx], self.labels[idx]
-        image = np.reshape(image, [1, 28, 28])
+        image = np.reshape(image, [28, 28])
+
+        if self.backend == 'pil':
+            image = Image.fromarray(image, mode='L')
+
         if self.transform is not None:
             image = self.transform(image)
+
+        if self.backend == 'pil':
+            return image, label.astype('int64')
+
         return image.astype(self.dtype), label.astype('int64')
 
     def __len__(self):

python/paddle/vision/datasets/voc2012.py

@@ -48,6 +48,10 @@ class VOC2012(Dataset):
         mode(str): 'train', 'valid' or 'test' mode. Default 'train'.
         download(bool): whether to download dataset automatically if
             :attr:`data_file` is not set. Default True
+        backend(str, optional): Specifies which type of image to be returned: 
+            PIL.Image or numpy.ndarray. Should be one of {'pil', 'cv2'}. 
+            If this option is not set, will get backend from ``paddle.vsion.get_image_backend`` ,
+            default backend is 'pil'. Default: None.
 
     Examples:
 
@@ -55,6 +59,7 @@ class VOC2012(Dataset):
 
             import paddle
             from paddle.vision.datasets import VOC2012
+            from paddle.vision.transforms import Normalize
 
             class SimpleNet(paddle.nn.Layer):
                 def __init__(self):
@@ -65,7 +70,10 @@ class VOC2012(Dataset):
 
             paddle.disable_static()
 
-            voc2012 = VOC2012(mode='train')
+            normalize = Normalize(mean=[0.5, 0.5, 0.5],
+                                  std=[0.5, 0.5, 0.5],
+                                  data_format='HWC')
+            voc2012 = VOC2012(mode='train', transform=normalize, backend='cv2')
 
             for i in range(10):
                 image, label= voc2012[i]
@@ -82,9 +90,19 @@ class VOC2012(Dataset):
                  data_file=None,
                  mode='train',
                  transform=None,
-                 download=True):
+                 download=True,
+                 backend=None):
         assert mode.lower() in ['train', 'valid', 'test'], \
             "mode should be 'train', 'valid' or 'test', but got {}".format(mode)
+
+        if backend is None:
+            backend = paddle.vision.get_image_backend()
+        if backend not in ['pil', 'cv2']:
+            raise ValueError(
+                "Expected backend are one of ['pil', 'cv2'], but got {}"
+                .format(backend))
+        self.backend = backend
+
         self.flag = MODE_FLAG_MAP[mode.lower()]
 
         self.data_file = data_file
@@ -126,12 +144,19 @@ class VOC2012(Dataset):
         label = self.data_tar.extractfile(self.name2mem[label_file]).read()
         data = Image.open(io.BytesIO(data))
         label = Image.open(io.BytesIO(label))
+
+        if self.backend == 'cv2':
             data = np.array(data)
             label = np.array(label)
+
         if self.transform is not None:
             data = self.transform(data)
+
+        if self.backend == 'cv2':
             return data.astype(self.dtype), label.astype(self.dtype)
 
+        return data, label
+
     def __len__(self):
         return len(self.data)
 

python/paddle/vision/transforms/transforms.py

@@ -686,6 +686,8 @@ class Transpose(BaseTransform):
         if F._is_pil_image(img):
             img = np.asarray(img)
 
+        if len(img.shape) == 2:
+            img = img[..., np.newaxis]
         return img.transpose(self.order)