GitLab

Summary:
Pull Request resolved: https://github.com/pytorch/vision/pull/3823

Uploaded FashionMNIST dataset to [manifold](https://www.internalfb.com/intern/network/manifold/?bucket=torchvision&path=tree%2Fdatasets) bucket `torchvision`. Any new dataset that needs to be added could be uploaded under `tree/datasets/<dataset_name>`.

Reviewed By: datumbox

Differential Revision: D28358470

fbshipit-source-id: 6f2282d3f1ce4b1416e962de8fb132896d4b2d76

* Added support for bicubic mode with AA

* Updated comment in the test

* Fix io imports and expose methods

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

* Partial implementation of SSDlite.

* Add normal init and BN hyperparams.

* Refactor to keep JIT happy

* Completed SSDlite.

* Fix lint

* Update todos

* Add expected file in repo.

* Use C4 expansion instead of C4 output.

* Change scales formula for Default Boxes.

* Add cosine annealing on trainer.

* Make T_max count epochs.

* Fix test and handle corner-case.

* Add support of support width_mult

* Add ssdlite presets.

* Change ReLU6, [-1,1] rescaling, backbone init & no pretraining.

* Use _reduced_tail=True.

* Add sync BN support.

* Adding the best config along with its weights and documentation.

* Make mean/std configurable.

* Fix not implemented for half exception

Co-authored-by: James Thewlis <james@unitary.ai>

* Fix inconsistent version pinning across yaml files.

* Pin pillow to >=5.3.0

* Pin pillow>=5.3.0 everywhere

* Refactor grid default boxes with torch.meshgrid

* Fix torch jit tracing

* Only doing the list multiplication once
Co-authored-by: Francisco Massa <fvsmassa@gmail.com>

* Make grid_default_box private as suggested
Co-authored-by: Vasilis Vryniotis <datumbox@users.noreply.github.com>

* Replace list multiplication with torch.repeat

* Move the clipping into _grid_default_boxes to accelerate
Co-authored-by: Francisco Massa <fvsmassa@gmail.com>
Co-authored-by: Vasilis Vryniotis <datumbox@users.noreply.github.com>

* WIP Added antialias option to transforms.functional.resize

* Updates according to the review

* Excluded these C++ files for iOS build

* Added support for mixed downsampling/upsampling

* Fixed heap overflow caused by explicit loop unrolling

* Applied PR review suggestions
- used pytest parametrize instead unittest
- cast to scalar_t ptr
- removed interpolate aa files for ios/android keeping original cmake version

* Early skeleton of API.

* Adding MultiFeatureMap and vgg16 backbone.

* Making vgg16 backbone same as paper.

* Making code generic to support all vggs.

* Moving vgg's extra layers a separate class + L2 scaling.

* Adding header vgg layers.

* Fix maxpool patching.

* Refactoring code to allow for support of different backbones & sizes:
- Skeleton for Default Boxes generator class
- Dynamic estimation of configuration when possible
- Addition of types

* Complete the implementation of DefaultBox generator.

* Replace randn with empty.

* Minor refactoring

* Making clamping between 0 and 1 optional.

* Change xywh to xyxy encoding.

* Adding parameters and reusing objects in constructor.

* Temporarily inherit from Retina to avoid dup code.

* Implement forward methods + temp workarounds to inherit from retina.

* Inherit more methods from retinanet.

* Fix type error.

* Add Regression loss.

* Fixing JIT issues.

* Change JIT workaround to minimize new code.

* Fixing initialization bug.

* Add classification loss.

* Update todos.

* Add weight loading support.

* Support SSD512.

* Change kernel_size to get output size 1x1

* Add xavier init and refactoring.

* Adding unit-tests and fixing JIT issues.

* Add a test for dbox generator.

* Remove unnecessary import.

* Workaround on GeneralizedRCNNTransform to support fixed size input.

* Remove unnecessary random calls from the test.

* Remove more rand calls from the test.

* change mapping and handling of empty labels

* Fix JIT warnings.

* Speed up loss.

* Convert 0-1 dboxes to original size.

* Fix warning.

* Fix tests.

* Update comments.

* Fixing minor bugs.

* Introduce a custom DBoxMatcher.

* Minor refactoring

* Move extra layer definition inside feature extractor.

* handle no bias on init.

* Remove fixed image size limitation

* Change initialization values for bias of classification head.

* Refactoring and update test file.

* Adding ResNet backbone.

* Minor refactoring.

* Remove inheritance of retina and general refactoring.

* SSD should fix the input size.

* Fixing messages and comments.

* Silently ignoring exception if test-only.

* Update comments.

* Update regression loss.

* Restore Xavier init everywhere, update the negative sampling method, change the clipping approach.

* Fixing tests.

* Refactor to move the losses from the Head to the SSD.

* Removing resnet50 ssd version.

* Adding support for best performing backbone and its config.

* Refactor and clean up the API.

* Fix lint

* Update todos and comments.

* Adding RandomHorizontalFlip and RandomIoUCrop transforms.

* Adding necessary checks to our tranforms.

* Adding RandomZoomOut.

* Adding RandomPhotometricDistort.

* Moving Detection transforms to references.

* Update presets

* fix lint

* leave compose and object

* Adding scaling for completeness.

* Adding params in the repr

* Remove unnecessary import.

* minor refactoring

* Remove unnecessary call.

* Give better names to DBox* classes

* Port num_anchors estimation in generator

* Remove rescaling and fix presets

* Add the ability to pass a custom head and refactoring.

* fix lint

* Fix unit-test

* Update todos.

* Change mean values.

* Change the default parameter of SSD to train the full VGG16 and remove the catch of exception for eval only.

* Adding documentation

* Adding weights and updating readmes.

* Update the model weights with a more performing model.

* Adding doc for head.

* Restore import.

* Refactored set_cell_anchors() in AnchorGenerator

* Addressed review comment

* Fixed test failure

* Keep consistency of ConvBNActivation

* Simplify using the Python3 idiom
Co-authored-by: Vasilis Vryniotis <datumbox@users.noreply.github.com>
Co-authored-by: Vasilis Vryniotis <datumbox@users.noreply.github.com>

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

* Add docs for Interpolation modes

* Update torchvision/transforms/functional.py
Co-authored-by: Nicolas Hug <contact@nicolas-hug.com>
Co-authored-by: Nicolas Hug <contact@nicolas-hug.com>
Co-authored-by: Vasilis Vryniotis <datumbox@users.noreply.github.com>

Co-authored-by: Nicolas Hug <nicolashug@fb.com>

* fix check for exceeded quota on Google Drive

* safeguard quota exceed check

* Edited roi transforms docs

* remove incorrect param desc

* Fixed confusion  about feature maps according to comment

* Remove pandas dependecy for CelebA dataset

* address PR comments

* Apply suggestions from code review
Co-authored-by: Philip Meier <github.pmeier@posteo.de>
Co-authored-by: Vasilis Vryniotis <datumbox@users.noreply.github.com>
Co-authored-by: Philip Meier <github.pmeier@posteo.de>

* Added KITTI dataset

* Addressed review comments

* Changed type of target to List[Dict] and corrected the data types of the returned values.

* Updated unit test to rely on ImageDatasetTestCase

* Added kitti to dataset documentation

* Cleaned up test and some minor changes

* Made data_url a string instead of a list

* Removed unnecessary try and print
Co-authored-by: Francisco Massa <fvsmassa@gmail.com>

* Make two methods as similar as possible.

* Introducing conditional fake casting.

* Change the casting mechanism.

* WIP

* clang

* docs

* extracted out common utils

* Use better quantization function and pass tensors as parameters

* proper dequantization

* Some tests

* Dequantization optimization, seems to gain a few ms

* clang-format

* again

* more correct test. Had to remove optimization although it almost works

* Also test aligned=True

* remove useless part

* more docs and comments

* Put back optimization with more robust test

* Added check for index upper bound

* avoid possible overflow

* Move common function into common.h

* oops

* scale=1,zero_point=0 makes more sense

* Force batch size of 1 to prevent any indexingbug

* format

* format again

* updated docstring

* put back description comment for pre_calc_bilinear_interpolate

* revert most changes to docstring as it's taken care of in another PR

* Fixed return docstrings

* Added some refs and corrected some parts

* more refs, and a note about dtypes
Co-authored-by: Francisco Massa <fvsmassa@gmail.com>

* Added defusedxml to parse untrusted XML data

* Added typecheck disable for defusedxml
Co-authored-by: Nicolas Hug <nicolashug@fb.com>

Co-authored-by: Nicolas Hug <contact@nicolas-hug.com>
Co-authored-by: Nicolas Hug <nicolashug@fb.com>