GitLab

Add document entry for datasets.Omniglot

* enable mypy for torchvision.datasets and ignore existing errors

* imagenet

* utils

* lint

Co-authored-by: dmitrysarov <d.shaulskiy@gmail.com>

* fix fill parameter for PIL pad

* add test

* fix

* lint

* [WIP] Unified input for T.RandomAffine

* Unified inputs for T.RandomAffine transformation

* Update transforms.py

* Updated docs of F.affine fillcolor

* Update transforms.py

* Update transform.py

* Update transform.py

* add torchscriptable adjust_gamma transform

https://github.com/pytorch/vision/issues/1375



* changes based on code-review

* Apply suggested change to add type hint

Required by mypy, even thought technically incorrect due to possible Image parameter. torchscript doesn't support a union based type hint.
Co-authored-by: vfdev <vfdev.5@gmail.com>
Co-authored-by: vfdev <vfdev.5@gmail.com>

* perform cyclic check for hue in test_rgb2hsv

Test fails for cases when hue=0 and hue=360. As hue is cyclic in nature, add cyclic logic for checking the max difference by taking the sin of the tensor and then comparing the max values.

* address linter issues

* [WIP] F.affine

* [WIP] F.affine + tests

* Unified input for F.affine

* Removed commented code

* Removed unused imports

* Add corruption cases

* Read jpeg headers until exhaustion

* Minor error correction

* Add test script

* Raise exception when image is truncated

* Add test

* Skip damaged_jpeg folder

* Compare against basename

* Remove unused test file

Incoherence is when affine transformation is 90 degrees rotation and output contains a zero line

* Remove linux distribution detection, search first on local paths for libraries

* Use not in

* Remove unused variables

* Fix wrong clamping in RoIAlign with aligned=True

* Fix silly mistake

* Bugfix pointed out during code-review

* Fixes Xiao's repro

* Ports nms to use full dispatcher

* Move HIPGuard to nms_cuda

* clang-format

* run models in test_models.py on GPU if available

* Francisco's comment, also disable cuda model tests to see if CPU alone still passes

* cuda tests now pass locally, although still not comparing to saved numerics

* add note for thing to ask francisco

* Allow cuda and cpu tests to share a data file

* ignore suffix if unneeded

* Skip autocast numerics checks for a few models

* Add roi_align test
Co-authored-by: Michael Carilli <mcarilli@nvidia.com>