New package: py-torchvision (#12175)

var/spack/repos/builtin/packages/py-torchvision/package.py

@@ -0,0 +1,33 @@
+# Copyright 2013-2019 Lawrence Livermore National Security, LLC and other
+# Spack Project Developers. See the top-level COPYRIGHT file for details.
+#
+# SPDX-License-Identifier: (Apache-2.0 OR MIT)
+
+from spack import *
+
+
+class PyTorchvision(PythonPackage):
+    """The torchvision package consists of popular datasets, model
+    architectures, and common image transformations for computer vision."""
+
+    homepage = "https://github.com/pytorch/vision"
+    url      = "https://github.com/pytorch/vision/archive/v0.3.0.tar.gz"
+
+    maintainers = ['adamjstewart']
+    import_modules = [
+        'torchvision', 'torchvision.datasets', 'torchvision.models',
+        'torchvision.transforms', 'torchvision.ops',
+        'torchvision.models.segmentation',
+        'torchvision.models.detection'
+    ]
+
+    version('0.3.0', sha256='c205f0618c268c6ed2f8abb869ef6eb83e5339c1336c243ad321a2f2a85195f0')
+
+    depends_on('py-setuptools', type='build')
+    depends_on('py-numpy', type=('build', 'run'))
+    depends_on('py-six', type=('build', 'run'))
+    depends_on('py-torch@1.1.0:', type=('build', 'run'))
+    depends_on('py-pillow@4.1.1:', type=('build', 'run'))  # or py-pillow-simd
+
+    # Many of the datasets require additional dependencies to use.
+    # These can be installed after the fact.