Add --use-v2 support to classification references (#7724)

Author: NicolasHug

references/classification/presets.py

@@ -1,8 +1,19 @@
 import torch
-from torchvision.transforms import autoaugment, transforms
 from torchvision.transforms.functional import InterpolationMode
 
 
+def get_module(use_v2):
+    # We need a protected import to avoid the V2 warning in case just V1 is used
+    if use_v2:
+        import torchvision.transforms.v2
+
+        return torchvision.transforms.v2
+    else:
+        import torchvision.transforms
+
+        return torchvision.transforms
+
+
 class ClassificationPresetTrain:
     def __init__(
         self,
@@ -17,41 +28,44 @@ def __init__(
         augmix_severity=3,
         random_erase_prob=0.0,
         backend="pil",
+        use_v2=False,
     ):
-        trans = []
+        module = get_module(use_v2)
+
+        transforms = []
         backend = backend.lower()
         if backend == "tensor":
-            trans.append(transforms.PILToTensor())
+            transforms.append(module.PILToTensor())
         elif backend != "pil":
             raise ValueError(f"backend can be 'tensor' or 'pil', but got {backend}")
 
-        trans.append(transforms.RandomResizedCrop(crop_size, interpolation=interpolation, antialias=True))
+        transforms.append(module.RandomResizedCrop(crop_size, interpolation=interpolation, antialias=True))
         if hflip_prob > 0:
-            trans.append(transforms.RandomHorizontalFlip(hflip_prob))
+            transforms.append(module.RandomHorizontalFlip(hflip_prob))
         if auto_augment_policy is not None:
             if auto_augment_policy == "ra":
-                trans.append(autoaugment.RandAugment(interpolation=interpolation, magnitude=ra_magnitude))
+                transforms.append(module.RandAugment(interpolation=interpolation, magnitude=ra_magnitude))
             elif auto_augment_policy == "ta_wide":
-                trans.append(autoaugment.TrivialAugmentWide(interpolation=interpolation))
+                transforms.append(module.TrivialAugmentWide(interpolation=interpolation))
             elif auto_augment_policy == "augmix":
-                trans.append(autoaugment.AugMix(interpolation=interpolation, severity=augmix_severity))
+                transforms.append(module.AugMix(interpolation=interpolation, severity=augmix_severity))
             else:
-                aa_policy = autoaugment.AutoAugmentPolicy(auto_augment_policy)
-                trans.append(autoaugment.AutoAugment(policy=aa_policy, interpolation=interpolation))
+                aa_policy = module.AutoAugmentPolicy(auto_augment_policy)
+                transforms.append(module.AutoAugment(policy=aa_policy, interpolation=interpolation))
 
         if backend == "pil":
-            trans.append(transforms.PILToTensor())
+            transforms.append(module.PILToTensor())
 
-        trans.extend(
+        transforms.extend(
             [
-                transforms.ConvertImageDtype(torch.float),
-                transforms.Normalize(mean=mean, std=std),
+                module.ConvertImageDtype(torch.float),
+                module.Normalize(mean=mean, std=std),
             ]
         )
         if random_erase_prob > 0:
-            trans.append(transforms.RandomErasing(p=random_erase_prob))
+            transforms.append(module.RandomErasing(p=random_erase_prob))
 
-        self.transforms = transforms.Compose(trans)
+        self.transforms = module.Compose(transforms)
 
     def __call__(self, img):
         return self.transforms(img)
@@ -67,28 +81,30 @@ def __init__(
         std=(0.229, 0.224, 0.225),
         interpolation=InterpolationMode.BILINEAR,
         backend="pil",
+        use_v2=False,
     ):
-        trans = []
+        module = get_module(use_v2)
+        transforms = []
         backend = backend.lower()
         if backend == "tensor":
-            trans.append(transforms.PILToTensor())
+            transforms.append(module.PILToTensor())
         elif backend != "pil":
             raise ValueError(f"backend can be 'tensor' or 'pil', but got {backend}")
 
-        trans += [
-            transforms.Resize(resize_size, interpolation=interpolation, antialias=True),
-            transforms.CenterCrop(crop_size),
+        transforms += [
+            module.Resize(resize_size, interpolation=interpolation, antialias=True),
+            module.CenterCrop(crop_size),
         ]
 
         if backend == "pil":
-            trans.append(transforms.PILToTensor())
+            transforms.append(module.PILToTensor())
 
-        trans += [
-            transforms.ConvertImageDtype(torch.float),
-            transforms.Normalize(mean=mean, std=std),
+        transforms += [
+            module.ConvertImageDtype(torch.float),
+            module.Normalize(mean=mean, std=std),
         ]
 
-        self.transforms = transforms.Compose(trans)
+        self.transforms = module.Compose(transforms)
 
     def __call__(self, img):
         return self.transforms(img)

references/classification/train.py

@@ -145,6 +145,7 @@ def load_data(traindir, valdir, args):
                 ra_magnitude=ra_magnitude,
                 augmix_severity=augmix_severity,
                 backend=args.backend,
+                use_v2=args.use_v2,
             ),
         )
         if args.cache_dataset:
@@ -172,6 +173,7 @@ def load_data(traindir, valdir, args):
                 resize_size=val_resize_size,
                 interpolation=interpolation,
                 backend=args.backend,
+                use_v2=args.use_v2,
             )
 
         dataset_test = torchvision.datasets.ImageFolder(
@@ -516,6 +518,7 @@ def get_args_parser(add_help=True):
     )
     parser.add_argument("--weights", default=None, type=str, help="the weights enum name to load")
     parser.add_argument("--backend", default="PIL", type=str.lower, help="PIL or tensor - case insensitive")
+    parser.add_argument("--use-v2", action="store_true", help="Use V2 transforms")
     return parser