Update CI on Multiweight branch to use the new weight download approach (#5628)

* port Pad to prototype transforms (#5621)

* port Pad to prototype transforms

* use literal

* Bump up LibTorchvision version number for Podspec to release Cocoapods (#5624)

Co-authored-by: Anton Thomma <anton@pri.co.nz>
Co-authored-by: Vasilis Vryniotis <datumbox@users.noreply.github.com>

* pre-download model weights in CI docs build (#5625)

* pre-download model weights in CI docs build

* move changes into template

* change docs image

* Regenerated config.yml

Co-authored-by: Philip Meier <github.pmeier@posteo.de>
Co-authored-by: Anton Thomma <11010310+thommaa@users.noreply.github.com>
Co-authored-by: Anton Thomma <anton@pri.co.nz>

Author: 4 people

.circleci/config.yml

@@ -174,6 +174,26 @@ commands:
       - store_test_results:
           path: test-results
 
+  download_model_weights:
+    parameters:
+      extract_roots:
+        type: string
+        default: "torchvision/models"
+      background:
+        type: boolean
+        default: true
+    steps:
+      - apt_install:
+          args: parallel wget
+          descr: Install download utilitites
+      - run:
+          name: Download model weights
+          background: << parameters.background >>
+          command: |
+            mkdir -p ~/.cache/torch/hub/checkpoints
+            python scripts/collect_model_urls.py << parameters.extract_roots >> \
+                | parallel -j0 'wget --no-verbose -O ~/.cache/torch/hub/checkpoints/`basename {}` {}\?source=ci'
+
 binary_common: &binary_common
   parameters:
     # Edit these defaults to do a release
@@ -354,14 +374,7 @@ jobs:
     resource_class: xlarge
     steps:
       - checkout
-      - run:
-          name: Download model weights
-          background: true
-          command: |
-            sudo apt update -qy && sudo apt install -qy parallel wget
-            mkdir -p ~/.cache/torch/hub/checkpoints
-            python scripts/collect_model_urls.py torchvision/models \
-                | parallel -j0 'wget --no-verbose -O ~/.cache/torch/hub/checkpoints/`basename {}` {}\?source=ci'
+      - download_model_weights
       - install_torchvision
       - run:
           name: Enable extended tests
@@ -1021,12 +1034,13 @@ jobs:
   build_docs:
     <<: *binary_common
     docker:
-      - image: "pytorch/manylinux-cuda100"
+      - image: circleci/python:3.7
     resource_class: 2xlarge+
     steps:
       - attach_workspace:
           at: ~/workspace
       - checkout
+      - download_model_weights
       - run:
           name: Setup
           command: .circleci/unittest/linux/scripts/setup_env.sh

.circleci/config.yml.in

@@ -1,8 +1,8 @@
-pytorch_version = '1.10.0'
+pytorch_version = '1.11.0'
 
 Pod::Spec.new do |s|
     s.name             = 'LibTorchvision'
-    s.version          = '0.11.1'
+    s.version          = '0.12.0'
     s.authors          = 'PyTorch Team'
     s.license          = { :type => 'BSD' }
     s.homepage         = 'https://github.com/pytorch/vision'

ios/LibTorchvision.podspec

@@ -2,21 +2,19 @@
 import re
 import sys
 
-MODEL_URL_PATTERN = re.compile(r"https://download[.]pytorch[.]org/models/.*?[.]pth")
+MODEL_URL_PATTERN = re.compile(r"https://download[.]pytorch[.]org/models/.+?[.]pth")
 
 
-def main(root):
+def main(*roots):
     model_urls = set()
-    for path in pathlib.Path(root).glob("**/*"):
-        if path.name.startswith("_") or not path.suffix == ".py":
-            continue
-
-        with open(path, "r") as file:
-            for line in file:
-                model_urls.update(MODEL_URL_PATTERN.findall(line))
+    for root in roots:
+        for path in pathlib.Path(root).rglob("*.py"):
+            with open(path, "r") as file:
+                for line in file:
+                    model_urls.update(MODEL_URL_PATTERN.findall(line))
 
     print("\n".join(sorted(model_urls)))
 
 
 if __name__ == "__main__":
-    main(sys.argv[1])
+    main(*sys.argv[1:])

scripts/collect_model_urls.py

@@ -71,6 +71,7 @@ class TestSmoke:
         transforms.CenterCrop([16, 16]),
         transforms.ConvertImageDtype(),
         transforms.RandomHorizontalFlip(),
+        transforms.Pad(5),
     )
     def test_common(self, transform, input):
         transform(input)

test/test_prototype_transforms.py

@@ -13,6 +13,7 @@
     TenCrop,
     BatchMultiCrop,
     RandomHorizontalFlip,
+    Pad,
     RandomZoomOut,
 )
 from ._meta import ConvertBoundingBoxFormat, ConvertImageDtype, ConvertImageColorSpace

torchvision/prototype/transforms/__init__.py

@@ -1,5 +1,6 @@
 import collections.abc
 import math
+import numbers
 import warnings
 from typing import Any, Dict, List, Union, Sequence, Tuple, cast
 
@@ -9,6 +10,7 @@
 from torchvision.prototype.transforms import Transform, functional as F
 from torchvision.transforms.functional import pil_to_tensor, InterpolationMode
 from torchvision.transforms.transforms import _setup_size, _interpolation_modes_from_int
+from typing_extensions import Literal
 
 from ._utils import query_image, get_image_dimensions, has_any, is_simple_tensor
 
@@ -272,42 +274,31 @@ def apply_recursively(obj: Any) -> Any:
         return apply_recursively(inputs if len(inputs) > 1 else inputs[0])
 
 
-class RandomZoomOut(Transform):
+class Pad(Transform):
     def __init__(
-        self, fill: Union[float, Sequence[float]] = 0.0, side_range: Tuple[float, float] = (1.0, 4.0), p: float = 0.5
+        self,
+        padding: Union[int, Sequence[int]],
+        fill: Union[float, Sequence[float]] = 0.0,
+        padding_mode: Literal["constant", "edge", "reflect", "symmetric"] = "constant",
     ) -> None:
         super().__init__()
+        if not isinstance(padding, (numbers.Number, tuple, list)):
+            raise TypeError("Got inappropriate padding arg")
 
-        if fill is None:
-            fill = 0.0
-        self.fill = fill
-
-        self.side_range = side_range
-        if side_range[0] < 1.0 or side_range[0] > side_range[1]:
-            raise ValueError(f"Invalid canvas side range provided {side_range}.")
-
-        self.p = p
-
-    def _get_params(self, sample: Any) -> Dict[str, Any]:
-        image = query_image(sample)
-        orig_c, orig_h, orig_w = get_image_dimensions(image)
-
-        r = self.side_range[0] + torch.rand(1) * (self.side_range[1] - self.side_range[0])
-        canvas_width = int(orig_w * r)
-        canvas_height = int(orig_h * r)
+        if not isinstance(fill, (numbers.Number, str, tuple, list)):
+            raise TypeError("Got inappropriate fill arg")
 
-        r = torch.rand(2)
-        left = int((canvas_width - orig_w) * r[0])
-        top = int((canvas_height - orig_h) * r[1])
-        right = canvas_width - (left + orig_w)
-        bottom = canvas_height - (top + orig_h)
-        padding = [left, top, right, bottom]
+        if padding_mode not in ["constant", "edge", "reflect", "symmetric"]:
+            raise ValueError("Padding mode should be either constant, edge, reflect or symmetric")
 
-        fill = self.fill
-        if not isinstance(fill, collections.abc.Sequence):
-            fill = [fill] * orig_c
+        if isinstance(padding, Sequence) and len(padding) not in [1, 2, 4]:
+            raise ValueError(
+                f"Padding must be an int or a 1, 2, or 4 element tuple, not a {len(padding)} element tuple"
+            )
 
-        return dict(padding=padding, fill=fill)
+        self.padding = padding
+        self.fill = fill
+        self.padding_mode = padding_mode
 
     def _transform(self, input: Any, params: Dict[str, Any]) -> Any:
         if isinstance(input, features.Image) or is_simple_tensor(input):
@@ -349,6 +340,48 @@ def _transform(self, input: Any, params: Dict[str, Any]) -> Any:
         else:
             return input
 
+
+class RandomZoomOut(Transform):
+    def __init__(
+        self, fill: Union[float, Sequence[float]] = 0.0, side_range: Tuple[float, float] = (1.0, 4.0), p: float = 0.5
+    ) -> None:
+        super().__init__()
+
+        if fill is None:
+            fill = 0.0
+        self.fill = fill
+
+        self.side_range = side_range
+        if side_range[0] < 1.0 or side_range[0] > side_range[1]:
+            raise ValueError(f"Invalid canvas side range provided {side_range}.")
+
+        self.p = p
+
+    def _get_params(self, sample: Any) -> Dict[str, Any]:
+        image = query_image(sample)
+        orig_c, orig_h, orig_w = get_image_dimensions(image)
+
+        r = self.side_range[0] + torch.rand(1) * (self.side_range[1] - self.side_range[0])
+        canvas_width = int(orig_w * r)
+        canvas_height = int(orig_h * r)
+
+        r = torch.rand(2)
+        left = int((canvas_width - orig_w) * r[0])
+        top = int((canvas_height - orig_h) * r[1])
+        right = canvas_width - (left + orig_w)
+        bottom = canvas_height - (top + orig_h)
+        padding = [left, top, right, bottom]
+
+        fill = self.fill
+        if not isinstance(fill, collections.abc.Sequence):
+            fill = [fill] * orig_c
+
+        return dict(padding=padding, fill=fill)
+
+    def _transform(self, input: Any, params: Dict[str, Any]) -> Any:
+        transform = Pad(**params, padding_mode="constant")
+        return transform(input)
+
     def forward(self, *inputs: Any) -> Any:
         sample = inputs if len(inputs) > 1 else inputs[0]
         if torch.rand(1) >= self.p: