GitLab

* Performance optimization on adjust_hue_image_tensor

* handle ints

* Inplace logical ops

* Remove unnecessary casting.

* Fix linter.

* Port elastic and minor cleanups

* Update torchvision/prototype/transforms/functional/_geometry.py
Co-authored-by: Philip Meier <github.pmeier@posteo.de>
Co-authored-by: Philip Meier <github.pmeier@posteo.de>

* Performance optimization for autocontrast

* Fixing tests

* Avoid double casting on adjust_contrast

* Handle properly ints.

* Speed up `adjust_sharpness_image_tensor`

* Add a comment

* [Nova] Caller for Conda M1 builds

* rename job

* add a signature consistency tests for v1 vs. v2 dispatchers

* temporarily increase test verbosity

* Revert "temporarily increase test verbosity"

This reverts commit 468c73f727af8dde8a2a3cc3063fd64923c50f63.

* fix test to allow annotation deviations

* fill <-> center for rotate

* ignore annotation changes for center / translate in rotate / affine

* add support for fine-grained tolerance settings

* fix test_cuda_vs_cpu

* add test for dispatcher kernel signature consistency

* add dispatcher feature signature consistency test

* fix error message
Co-authored-by: Vasilis Vryniotis <datumbox@users.noreply.github.com>

* [proto] small optim for perspective op on images, reverted concat trick on bboxes

* revert unrelated changes

* PR review updates

* PR review change

* fix ONNX tests

* fix error trigger

* remove relaxing condition

* remove tolerate_small_differences

* add KernelInfo's for {five, ten}_crop_video

* add video kernels to dispatcher tests

This reverts commit 2e833520.

* Restore BC on perspective

* Fixes linter

* Fixing tests.

* Apply code-review changes.

* Pleasing mypy.

* Revert named parameters.

* [proto] Small optims for elastic op on bboxes

* More inplace ops according to the review

* Create grid on device directly. This should be faster

* PR Review update. Apply ceil on float input

* replace tensor division with scalar division and tensor multiplication

* fix consistency test tolerances

* fix prototype kernels

* fix stable kernels

* fix tests

* make test more robust

* improve invert for signed integers

* improve invert

* fix posterize

* Revert "assume that integer images are [0, 255] in equalize (#6859)"

This reverts commit 436ff9a5

.

* fix solarize in AA

* fix resize

* Revert "fix resize"

This reverts commit 5f33f4aa73e098237650b794157ec9335d964152.

* add comment to float max value
Co-authored-by: Vasilis Vryniotis <datumbox@users.noreply.github.com>

* remove unnecessary changes from pad_image_tensor

* cleanup

* fix fill=None workaround

* address review comments

* remove more xfails

* Bbox resize optimization

* Other (untested) optimizations on `_affine_bounding_box_xyxy` and `elastic_bounding_box`.

* fix conflict

* Reverting changes on elastic

* revert one more change

* Further improvement

* [proto] Speed-up crop on bboxes and tests

* Fix linter

* Update _geometry.py

* Fixed device issue

* Revert changes in test/prototype_transforms_kernel_infos.py

* Fixed failing correctness tests

* [proto] Optimized functional pad op for bboxes + tests

* Renamed copy-pasted variable name

* [proto] Small optims for perspective bboxes op

* [proto] Speed-up crop on bboxes and tests

* Fix linter

* Update _geometry.py

* Fixed device issue

* Revert changes in test/prototype_transforms_kernel_infos.py

* Fixed failing correctness tests

* [proto] Optimized functional pad op for bboxes + tests

* Renamed copy-pasted variable name

* Code update

* Fixes according to the review

* Ported `resize`

* Align with previous behaviour

* Update torchvision/prototype/transforms/functional/_geometry.py
Co-authored-by: Philip Meier <github.pmeier@posteo.de>

* Moving input verification on top of method.
Co-authored-by: Philip Meier <github.pmeier@posteo.de>

* [proto] Speed-up crop on bboxes and tests

* Fix linter

* Update _geometry.py

* Fixed device issue

* Revert changes in test/prototype_transforms_kernel_infos.py

* Fixed failing correctness tests

* Refactor gaussian_blur

* Add conditional reshape

* Further refactoring

* Remove unused import.

* make weights deepcopyable

* add test

* test enum member instead of whole enum

* Reducing unnecessary method calls

* Optimize pad branch.

* Remove unnecessary call to crop_image_tensor

* Fix linter
Co-authored-by: vfdev <vfdev.5@gmail.com>

* [proto][tests] Added ref functions for h/v flips

* Better dtype handling in reference_affine_bounding_box_helper

* [proto] Speed-up h/v bboxes flip ops

* Use more inplace ops

* Removed _old methods

* Fixed jit issue using a bit slower version

* Adjust solarize threshold on input type

* Handle PIL images

* minor refactor

* Fix linter