Merge branch 'main' into please_dont_modify_this_branch_unless_you_are_just_merging_with_main__

NicolasHug · web-flow · commit 29f078d56a7f · 2024-11-06T10:12:57.000Z
diff --git a/.github/workflows/update-viablestrict.yml b/.github/workflows/update-viablestrict.yml
diff --git a/setup.py b/setup.py
@@ -127,7 +127,7 @@ def get_macros_and_flags():
             if NVCC_FLAGS is None:
                 nvcc_flags = []
             else:
-                nvcc_flags = nvcc_flags.split(" ")
+                nvcc_flags = NVCC_FLAGS.split(" ")
         extra_compile_args["nvcc"] = nvcc_flags
 
     if sys.platform == "win32":
diff --git a/torchvision/datasets/__init__.py b/torchvision/datasets/__init__.py
@@ -72,6 +72,7 @@
     "QMNIST",
     "MNIST",
     "KMNIST",
+    "MovingMNIST",
     "StanfordCars",
     "STL10",
     "SUN397",
diff --git a/torchvision/datasets/_stereo_matching.py b/torchvision/datasets/_stereo_matching.py
@@ -588,7 +588,6 @@ def _download_dataset(self, root: Union[str, Path]) -> None:
                 for calibration in ["perfect", "imperfect"]:
                     scene_name = f"{split_scene}-{calibration}"
                     scene_url = f"{base_url}/{scene_name}.zip"
-                    print(f"Downloading {scene_url}")
                     # download the scene only if it doesn't exist
                     if not (split_root / scene_name).exists():
                         download_and_extract_archive(
diff --git a/torchvision/datasets/caltech.py b/torchvision/datasets/caltech.py
@@ -130,7 +130,6 @@ def __len__(self) -> int:
 
     def download(self) -> None:
         if self._check_integrity():
-            print("Files already downloaded and verified")
             return
 
         download_and_extract_archive(
@@ -231,7 +230,6 @@ def __len__(self) -> int:
 
     def download(self) -> None:
         if self._check_integrity():
-            print("Files already downloaded and verified")
             return
 
         download_and_extract_archive(
diff --git a/torchvision/datasets/celeba.py b/torchvision/datasets/celeba.py
@@ -105,7 +105,7 @@ def __init__(
         if mask == slice(None):  # if split == "all"
             self.filename = splits.index
         else:
-            self.filename = [splits.index[i] for i in torch.squeeze(torch.nonzero(mask))]
+            self.filename = [splits.index[i] for i in torch.squeeze(torch.nonzero(mask))]  # type: ignore[arg-type]
         self.identity = identity.data[mask]
         self.bbox = bbox.data[mask]
         self.landmarks_align = landmarks_align.data[mask]
@@ -148,7 +148,6 @@ def _check_integrity(self) -> bool:
 
     def download(self) -> None:
         if self._check_integrity():
-            print("Files already downloaded and verified")
             return
 
         for (file_id, md5, filename) in self.file_list:
diff --git a/torchvision/datasets/cifar.py b/torchvision/datasets/cifar.py
@@ -135,7 +135,6 @@ def _check_integrity(self) -> bool:
 
     def download(self) -> None:
         if self._check_integrity():
-            print("Files already downloaded and verified")
             return
         download_and_extract_archive(self.url, self.root, filename=self.filename, md5=self.tgz_md5)
 
diff --git a/torchvision/datasets/cityscapes.py b/torchvision/datasets/cityscapes.py
@@ -192,7 +192,7 @@ def __getitem__(self, index: int) -> Tuple[Any, Any]:
 
             targets.append(target)
 
-        target = tuple(targets) if len(targets) > 1 else targets[0]
+        target = tuple(targets) if len(targets) > 1 else targets[0]  # type: ignore[assignment]
 
         if self.transforms is not None:
             image, target = self.transforms(image, target)
diff --git a/torchvision/datasets/imagenette.py b/torchvision/datasets/imagenette.py
@@ -81,10 +81,7 @@ def _check_exists(self) -> bool:
 
     def _download(self):
         if self._check_exists():
-            raise RuntimeError(
-                f"The directory {self._size_root} already exists. "
-                f"If you want to re-download or re-extract the images, delete the directory."
-            )
+            return
 
         download_and_extract_archive(self._url, self.root, md5=self._md5)
 
diff --git a/torchvision/datasets/inaturalist.py b/torchvision/datasets/inaturalist.py
@@ -239,4 +239,3 @@ def download(self) -> None:
         if not os.path.exists(orig_dir_name):
             raise RuntimeError(f"Unable to find downloaded files at {orig_dir_name}")
         os.rename(orig_dir_name, self.root)
-        print(f"Dataset version '{self.version}' has been downloaded and prepared for use")
diff --git a/torchvision/datasets/kinetics.py b/torchvision/datasets/kinetics.py
@@ -1,6 +1,5 @@
 import csv
 import os
-import time
 import urllib
 from functools import partial
 from multiprocessing import Pool
@@ -121,7 +120,6 @@ def __init__(
         self._legacy = _legacy
 
         if _legacy:
-            print("Using legacy structure")
             self.split_folder = root
             self.split = "unknown"
             output_format = "THWC"
@@ -157,14 +155,8 @@ def __init__(
 
     def download_and_process_videos(self) -> None:
         """Downloads all the videos to the _root_ folder in the expected format."""
-        tic = time.time()
         self._download_videos()
-        toc = time.time()
-        print("Elapsed time for downloading in mins ", (toc - tic) / 60)
         self._make_ds_structure()
-        toc2 = time.time()
-        print("Elapsed time for processing in mins ", (toc2 - toc) / 60)
-        print("Elapsed time overall in mins ", (toc2 - tic) / 60)
 
     def _download_videos(self) -> None:
         """download tarballs containing the video to "tars" folder and extract them into the _split_ folder where
diff --git a/torchvision/datasets/lfw.py b/torchvision/datasets/lfw.py
@@ -74,7 +74,6 @@ def _check_integrity(self) -> bool:
 
     def download(self) -> None:
         if self._check_integrity():
-            print("Files already downloaded and verified")
             return
         url = f"{self.download_url_prefix}{self.filename}"
         download_and_extract_archive(url, self.root, filename=self.filename, md5=self.md5)
diff --git a/torchvision/datasets/mnist.py b/torchvision/datasets/mnist.py
@@ -181,19 +181,20 @@ def download(self) -> None:
 
         # download files
         for filename, md5 in self.resources:
+            errors = []
             for mirror in self.mirrors:
                 url = f"{mirror}{filename}"
                 try:
-                    print(f"Downloading {url}")
                     download_and_extract_archive(url, download_root=self.raw_folder, filename=filename, md5=md5)
-                except URLError as error:
-                    print(f"Failed to download (trying next):\n{error}")
+                except URLError as e:
+                    errors.append(e)
                     continue
-                finally:
-                    print()
                 break
             else:
-                raise RuntimeError(f"Error downloading {filename}")
+                s = f"Error downloading {filename}:\n"
+                for mirror, err in zip(self.mirrors, errors):
+                    s += f"Tried {mirror}, got:\n{str(err)}\n"
+                raise RuntimeError(s)
 
     def extra_repr(self) -> str:
         split = "Train" if self.train is True else "Test"
diff --git a/torchvision/datasets/omniglot.py b/torchvision/datasets/omniglot.py
@@ -91,7 +91,6 @@ def _check_integrity(self) -> bool:
 
     def download(self) -> None:
         if self._check_integrity():
-            print("Files already downloaded and verified")
             return
 
         filename = self._get_target_folder()
diff --git a/torchvision/datasets/phototour.py b/torchvision/datasets/phototour.py
@@ -145,7 +145,6 @@ def _check_downloaded(self) -> bool:
 
     def download(self) -> None:
         if self._check_datafile_exists():
-            print(f"# Found cached data {self.data_file}")
             return
 
         if not self._check_downloaded():
@@ -157,8 +156,6 @@ def download(self) -> None:
 
             download_url(url, self.root, filename, md5)
 
-            print(f"# Extracting data {self.data_down}\n")
-
             import zipfile
 
             with zipfile.ZipFile(fpath, "r") as z:
@@ -168,7 +165,6 @@ def download(self) -> None:
 
     def cache(self) -> None:
         # process and save as torch files
-        print(f"# Caching data {self.data_file}")
 
         dataset = (
             read_image_file(self.data_dir, self.image_ext, self.lens[self.name]),
diff --git a/torchvision/datasets/sbu.py b/torchvision/datasets/sbu.py
@@ -93,7 +93,6 @@ def download(self) -> None:
         """Download and extract the tarball, and download each individual photo."""
 
         if self._check_integrity():
-            print("Files already downloaded and verified")
             return
 
         download_and_extract_archive(self.url, self.root, self.root, self.filename, self.md5_checksum)
diff --git a/torchvision/datasets/semeion.py b/torchvision/datasets/semeion.py
@@ -85,7 +85,6 @@ def _check_integrity(self) -> bool:
 
     def download(self) -> None:
         if self._check_integrity():
-            print("Files already downloaded and verified")
             return
 
         root = self.root
diff --git a/torchvision/datasets/stanford_cars.py b/torchvision/datasets/stanford_cars.py
@@ -15,6 +15,7 @@ class StanfordCars(VisionDataset):
     has been split roughly in a 50-50 split
 
     The original URL is https://ai.stanford.edu/~jkrause/cars/car_dataset.html, but it is broken.
+    Follow the instructions in ``download`` argument to obtain and use the dataset offline.
 
     .. note::
 
@@ -29,8 +30,12 @@ class StanfordCars(VisionDataset):
             target and transforms it.
         download (bool, optional): This parameter exists for backward compatibility but it does not
             download the dataset, since the original URL is not available anymore. The dataset
-            seems to be available on Kaggle so you can try to manually download it using
-            `these instructions <https://github.com/pytorch/vision/issues/7545#issuecomment-1631441616>`_.
+            seems to be available on Kaggle so you can try to manually download and configure it using
+            `these instructions <https://github.com/pytorch/vision/issues/7545#issuecomment-1631441616>`_,
+            or use an integrated
+            `dataset on Kaggle <https://github.com/pytorch/vision/issues/7545#issuecomment-2282674373>`_.
+            In both cases, first download and configure the dataset locally, and use the dataset with
+            ``"download=False"``.
     """
 
     def __init__(
diff --git a/torchvision/datasets/stl10.py b/torchvision/datasets/stl10.py
@@ -154,7 +154,6 @@ def _check_integrity(self) -> bool:
 
     def download(self) -> None:
         if self._check_integrity():
-            print("Files already downloaded and verified")
             return
         download_and_extract_archive(self.url, self.root, filename=self.filename, md5=self.tgz_md5)
         self._check_integrity()
diff --git a/torchvision/datasets/utils.py b/torchvision/datasets/utils.py
@@ -112,7 +112,6 @@ def download_url(
 
     # check if file is already present locally
     if check_integrity(fpath, md5):
-        print("Using downloaded and verified file: " + fpath)
         return
 
     if _is_remote_location_available():
@@ -128,12 +127,10 @@ def download_url(
 
         # download the file
         try:
-            print("Downloading " + url + " to " + fpath)
             _urlretrieve(url, fpath)
         except (urllib.error.URLError, OSError) as e:  # type: ignore[attr-defined]
             if url[:5] == "https":
                 url = url.replace("https:", "http:")
-                print("Failed download. Trying https -> http instead. Downloading " + url + " to " + fpath)
                 _urlretrieve(url, fpath)
             else:
                 raise e
@@ -204,7 +201,6 @@ def download_file_from_google_drive(
     os.makedirs(root, exist_ok=True)
 
     if check_integrity(fpath, md5):
-        print(f"Using downloaded {'and verified ' if md5 else ''}file: {fpath}")
         return
 
     gdown.download(id=file_id, output=fpath, quiet=False, user_agent=USER_AGENT)
@@ -395,7 +391,6 @@ def download_and_extract_archive(
     download_url(url, download_root, filename, md5)
 
     archive = os.path.join(download_root, filename)
-    print(f"Extracting {archive} to {extract_root}")
     extract_archive(archive, extract_root, remove_finished)
 
 
diff --git a/torchvision/datasets/widerface.py b/torchvision/datasets/widerface.py
@@ -182,7 +182,6 @@ def _check_integrity(self) -> bool:
 
     def download(self) -> None:
         if self._check_integrity():
-            print("Files already downloaded and verified")
             return
 
         # download and extract image data
diff --git a/torchvision/io/video.py b/torchvision/io/video.py
@@ -62,7 +62,12 @@ def write_video(
     audio_options: Optional[Dict[str, Any]] = None,
 ) -> None:
     """
-    Writes a 4d tensor in [T, H, W, C] format in a video file
+    Writes a 4d tensor in [T, H, W, C] format in a video file.
+
+    This function relies on PyAV (therefore, ultimately FFmpeg) to encode
+    videos, you can get more fine-grained control by referring to the other
+    options at your disposal within `the FFMpeg wiki
+    <http://trac.ffmpeg.org/wiki#Encoding>`_.
 
     .. warning::
 
@@ -78,12 +83,25 @@ def write_video(
             as a uint8 tensor in [T, H, W, C] format
         fps (Number): video frames per second
         video_codec (str): the name of the video codec, i.e. "libx264", "h264", etc.
-        options (Dict): dictionary containing options to be passed into the PyAV video stream
+        options (Dict): dictionary containing options to be passed into the PyAV video stream.
+            The list of options is codec-dependent and can all
+            be found from `the FFMpeg wiki <http://trac.ffmpeg.org/wiki#Encoding>`_.
         audio_array (Tensor[C, N]): tensor containing the audio, where C is the number of channels
             and N is the number of samples
         audio_fps (Number): audio sample rate, typically 44100 or 48000
         audio_codec (str): the name of the audio codec, i.e. "mp3", "aac", etc.
-        audio_options (Dict): dictionary containing options to be passed into the PyAV audio stream
+        audio_options (Dict): dictionary containing options to be passed into the PyAV audio stream.
+            The list of options is codec-dependent and can all
+            be found from `the FFMpeg wiki <http://trac.ffmpeg.org/wiki#Encoding>`_.
+
+    Examples::
+        >>> # Creating libx264 video with CRF 17, for visually lossless footage:
+        >>>
+        >>> from torchvision.io import write_video
+        >>> # 1000 frames of 100x100, 3-channel image.
+        >>> vid = torch.randn(1000, 100, 100, 3, dtype = torch.uint8)
+        >>> write_video("video.mp4", options = {"crf": "17"})
+
     """
     if not torch.jit.is_scripting() and not torch.jit.is_tracing():
         _log_api_usage_once(write_video)
diff --git a/torchvision/transforms/_functional_pil.py b/torchvision/transforms/_functional_pil.py
@@ -110,7 +110,7 @@ def adjust_hue(img: Image.Image, hue_factor: float) -> Image.Image:
 
     np_h = np.array(h, dtype=np.uint8)
     # This will over/underflow, as desired
-    np_h += np.array(hue_factor * 255).astype(np.uint8)
+    np_h += np.int32(hue_factor * 255).astype(np.uint8)
 
     h = Image.fromarray(np_h, "L")
 
