Use non deprecated resnet weights to compute camelyon16 features

flamby/datasets/fed_camelyon16/dataset_creation_scripts/tiling_slides.py

@@ -16,7 +16,7 @@
 from openslide import open_slide
 from torch.nn import Identity
 from torch.utils.data import DataLoader, Dataset, IterableDataset
-from torchvision.transforms import Compose, Normalize, ToTensor
+from torchvision.transforms import Compose, ToTensor
 from tqdm import tqdm
 
 from flamby.utils import read_config, write_value_in_config
@@ -101,7 +101,9 @@ def save_dict_to_csv(dict_arg, file_name):
     df.to_csv(file_name, index=False)
 
 
-def main(batch_size, num_workers_torch, tile_from_scratch, remove_big_tiff, output_path):
+def main(
+    batch_size, num_workers_torch, tile_from_scratch, remove_big_tiff, output_path
+):
     """Function tiling slides that have been downloaded using download.py.
 
     Parameters
@@ -142,8 +144,8 @@ def main(batch_size, num_workers_torch, tile_from_scratch, remove_big_tiff, outp
             debug = True
         else:
             raise ValueError(
-                "The dataset was not downloaded in normal or debug mode, \
-                    please run the download script beforehand"
+                "The dataset was not downloaded in normal or debug mode,"
+                " please run the download script beforehand"
             )
 
     if debug:
@@ -175,8 +177,17 @@ def main(batch_size, num_workers_torch, tile_from_scratch, remove_big_tiff, outp
         suffix=".png",
         tissue_percent=60,
     )
-    net = models.resnet50(pretrained=True)
+
+    # syntax with pretrained=True is now deprecated
+    # https://pytorch.org/vision/stable/models.html#initializing-pre-trained-models
+    resnet_weights = models.ResNet50_Weights.IMAGENET1K_V1
+    net = models.resnet50(weights=resnet_weights)
     net.fc = Identity()
+
+    # IMAGENET preprocessing of images scaled between 0. and 1. with ToTensor
+    resnet_transform = lambda img: resnet_weights.transforms(antialias=True)(img)
+    transform = Compose([ToTensor(), resnet_transform])
+
     # Probably unnecessary still it feels safer and might speed up computations
     for param in net.parameters():
         param.requires_grad = False
@@ -188,13 +199,10 @@ def main(batch_size, num_workers_torch, tile_from_scratch, remove_big_tiff, outp
         net = net.cuda()
     else:
         print(
-            "Consider using an environment with GPU otherwise the process will \
-                take weeks."
+            "Consider using an environment with GPU otherwise the process will "
+            "take weeks."
         )
-    # IMAGENET preprocessing of images scaled between 0. and 1. with ToTensor
-    transform = Compose(
-        [ToTensor(), Normalize(mean=[0.485, 0.456, 0.406], std=[0.229, 0.224, 0.225])]
-    )
+
     path_to_coords_file = os.path.join(
         Path(os.path.realpath(__file__)).parent.resolve(),
         "tiling_coordinates_camelyon16.csv",
@@ -291,7 +299,7 @@ def main(batch_size, num_workers_torch, tile_from_scratch, remove_big_tiff, outp
     if output_path is not None:
         write_value_in_config(config_file, "dataset_path", output_path)
 
-    if args.remove_big_tiff:
+    if remove_big_tiff:
         print("Removing all slides")
         for slide in slides_paths:
             os.remove(slide)
@@ -319,8 +327,10 @@ def main(batch_size, num_workers_torch, tile_from_scratch, remove_big_tiff, outp
     parser.add_argument(
         "--remove-big-tiff",
         action="store_true",
-        help="Whether or not to remove the original slides images that take \
-            up to 800G, after computing the features using them.",
+        help=(
+            "Whether or not to remove the original slides images that take "
+            "up to 800G, after computing the features using them."
+        ),
     )
     parser.add_argument(
         "--output-path", type=str, help="The path where to store the tiles"