GitLab

Co-authored-by: vfdev <vfdev.5@gmail.com>
Co-authored-by: Francisco Massa <fvsmassa@gmail.com>

* deprecate _video modules

* flake8
Co-authored-by: Vasilis Vryniotis <datumbox@users.noreply.github.com>

* always use default config as base

* fix test_all_configs decorator

* lint

* add a utility function to create a random string

* move output check of inject_fake_data to dedicated method

* always disable download and extract functionality

* make UsageError an Exception rather than RuntimeError

* separate fake data injection and dataset args handling

* adapt tests for Coco

* fix Coco implementation

* add documentation

* fix VideoDatasetTestCase

* adapt UCF101 tests

* cleanup

* allow FileNotFoundError for test without fake data

* Revert "fix Coco implementation"

This reverts commit e2b693881654b3e2462a73e6d22bb01c1b738f8a.

* lint

* fix UCF101 tests

* enable default frames per clips for video test cases

* add tests for UCF101

* remove old tests as well as fake data generation

* better explain frames_per_clip overriding

* lint
Co-authored-by: Francisco Massa <fvsmassa@gmail.com>

Summary:
Pull Request resolved: https://github.com/pytorch/vision/pull/3422

In CUDA-10.1 or below it is a no-op, but few of the tests will start to fail if update to 10.2 or newer, for example:
```
torchvision/transforms/transforms.py", line 1043, in forward
    transformed_tensor = torch.mm(flat_tensor, self.transformation_matrix)
RuntimeError: Deterministic behavior was enabled with either `torch.use_deterministic_algorithms(True)` or `at::Context::setDeterministicAlgorithms(true)`, but this operation is not deterministic because it uses CuBLAS and you have CUDA >= 10.2. To enable deterministic behavior in this case, you must set an environment variable before running your PyTorch application: CUBLAS_WORKSPACE_CONFIG=:4096:8 or CUBLAS_WORKSPACE_CONFIG=:16:8. For more information, go to https://docs.nvidia.com/cuda/cublas/index.html#cublasApi_reproducibility


```

Reviewed By: meyering

Differential Revision: D26563045

fbshipit-source-id: 65e5fabc8cdb471a72f931df55389ed7c0063e54
Co-authored-by: Nikita Shulga <nshulga@fb.com>

Co-authored-by: Francisco Massa <fvsmassa@gmail.com>

* use common download utils in VOC and SBDataset

* add tests for VOC

* use common base class for VOC datasets

* remove old voc test and fake data generation
Co-authored-by: Francisco Massa <fvsmassa@gmail.com>

Co-authored-by: Francisco Massa <fvsmassa@gmail.com>

Co-authored-by: Francisco Massa <fvsmassa@gmail.com>

* add base class for datasets tests

* add better type hints

* add documentation to subclasses

* add utility functions to create files / folders of random images and videos

* fix imports

* remove class properties

* fix smoke test

* fix type hints

* fix random size generation

* add Caltech256 as example

* add utility function to create grid of combinations

* add CIFAR100? as example

* lint

* add missing import

* improve documentation

* create 1 frame videos by default

* remove obsolete check

* return path of files created with utility functions

* [test] close PIL file handles before deletion

* fix video folder creation

* generalize file handle closing

* fix lazy imports

* add test for transforms

* fix explanation comment

* lint

* force load opened PIL images

* lint

* copy default config to avoid inplace modification

* enable additional arg forwarding

Fixes #3393

* Replace type T with accumulator.

* Upcast tensors of box ops to avoid overflow in multiplications.

Summary:
In opt mode, the linker could strip out the symbols related to the torchvision ops if they had not been explicitly requested.

Adding link-whole is one way of fixing this.

Reviewed By: datumbox, ppwwyyxx

Differential Revision: D26249713

fbshipit-source-id: 4e55e2ebb483922fe308fd8b83456439bde728c9

* Fix test flakiness caused by rounding.

* Update test/test_transforms.py

* Styles
Co-authored-by: Nicolas Hug <contact@nicolas-hug.com>

Summary:
Pull Request resolved: https://github.com/pytorch/vision/pull/3370

Pull Request resolved: https://github.com/pytorch/vision/pull/3361

Convert pytorch_gpu to use GPU/RE

Reviewed By: datumbox

Differential Revision: D26264102

fbshipit-source-id: b45179215ecb1142f6468386c4fee1759dd82f0c

(cherry picked from commit 2f40a483d73018ae6e1488a484c5927f2b309969)
Signed-off-by: Eli Uriegas <eliuriegas@fb.com>

* On MSVC, instruct the linker not to drop the _register_ops symbol

* Remove workaround for ops registration on windows
Co-authored-by: Francisco Massa <fvsmassa@gmail.com>

* remove generated dataset folders after download tests

* revert unnecessary style changes

* aspect ratio must be a sampling from log scale.
reference from: https://github.com/pytorch/vision/blob/8317295c1d272e0ba7b2ce31e3fd2c048235fc73/torchvision/transforms/transforms.py#L833-L836



* add random erasing unittest code

* Increased threshold for difference in sampling rate

* move outside of the loop

* log_ratio move outside of the loop. RandomResizedCrop also
Co-authored-by: Vasilis Vryniotis <datumbox@users.noreply.github.com>

Summary:
This diff adds a new target to torchvision which enables users to use torchvision ops from C++.

For now, the `cpp_library` is not used by the `python_cpp_library`. We should instead refactor the logic in torchvision to directly use `cpp_library` instead.

There is currently an inconsistency between fbcode and OSS users. OSS users can import torchvision via
```
#include <torchvision/vision.h>
```
while fbcode users need to do
```
#include <torchvision/csrc/vision.h>
```
It would be good to fix this discrepancy in the future.

I didn't directly use `test_frcnn_tracing.cpp` due to complications for getting the `.pt` file in a way that works for both OSS and fbcode, so instead we added a self-contained test that should validate that the torchvision ops are properly registered and visible to JIT

Reviewed By: datumbox

Differential Revision: D26225669

fbshipit-source-id: 5dd9fb98dd58e854f95806e4860d02f54fc04ea4
Co-authored-by: Francisco Massa <fmassa@fb.com>

* extract some functionality from places365 fakedata for common use

* add a common DatasetTestcase

* add fakedata generation and tests for STL10

* lint
Co-authored-by: Francisco Massa <fvsmassa@gmail.com>

* make timeout a parameter

* check redirects
Co-authored-by: Francisco Massa <fvsmassa@gmail.com>

* kmnist

* emnist

* qmnist

* omniglot

* phototour

* sbdataset

* sbu

* semeion

* stl10

* svhn

* usps

* cifar100

* enable download logging for google drive

* celeba

* widerface

* lint

* add timeout logic

* lint

* debug CI connection to problematic server

* set timeout for ping

* [ci skip] remove ping

* revert debugging

* disable requests to problematic server

* re-enable all other tests
Co-authored-by: Vasilis Vryniotis <datumbox@users.noreply.github.com>

* Check if the file is located on Google Drive

download_file() detects if the provided URL contains a link to Google
Drive and passes the request through the download_file_from_google_drive()
if it is the case.

* remove lazy re import

* add guard clause

* use urlparse

* add tests

* Fixing lint.

* make id matching more safe

* Fixing type check.

* Remove AnyStr.
Co-authored-by: Philip Meier <github.pmeier@posteo.de>
Co-authored-by: Vasilis Vryniotis <vvryniotis@fb.com>

* cleanup

* mnist

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

* Renamed original method to test center crop

* Added test method, docs and added padding when imgsize < cropsize.

* BugFix - keep odd_crop_size odd

* Do not crop when image size after padding matches crop size; updated test.
Co-authored-by: Vasilis Vryniotis <datumbox@users.noreply.github.com>

* adds fill paramter

* small doc edit

* Change fill to bool

* add filled

* fix the bugs

* Fixes bugs

* adds test with fill param

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

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

* initial fix

* fill=0

* docstrings

* fill type check

* fill type check

* set None to zero

* unit tests

* set instead of NotImplemented

* fix W293
Co-authored-by: vfdev <vfdev.5@gmail.com>
Co-authored-by: Vasilis Vryniotis <datumbox@users.noreply.github.com>

* Making _segm_resnet() generic and reusable.

* Adding fcn and deeplabv3 directly on mobilenetv3 backbone.

* Adding tests for segmentation models.

* Rename is_strided with _is_cn.

* Add dilation support on MobileNetV3 for Segmentation.

* Add Lite R-ASPP with MobileNetV3 backbone.

* Add pretrained model weights.

* Removing model fcn_mobilenet_v3_large.

* Adding docs and imports.

* Fixing typo and readme.

* Modify segmentation tests compare against expected values.

* Exclude flaky autocast tests.
Co-authored-by: Francisco Massa <fvsmassa@gmail.com>

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

* Tag fasterrcnn mobilenetv3 model with 320, add new inference config that makes it 2x faster sacrificing a bit of mAP.

* Add a high resolution fasterrcnn mobilenetv3 model.

* Update tests and expected values.