GitLab

* diff_iou_rotated is working

* add test; fix lint

* fix lint for test

* disable cpu build

* refactor files structure

* fix comments

* remove extra .repeat()

* add comment

* fix j-1 bug; update doc

* fix clang lint

* update docstrings

* fix comments

* fix comments

* Make sure padding size is not negative

If Resize maintains aspect ratio, could be the case that image shape is bigger than resize value.

* Test pad shape smaller than image shape

* Fix lint

* Remove whitespaces

* add gather_keep_from_mask_parallize

* remove unused cache

* move syncthread

* remove unused comment

* add more comments, rename the kernel and variable

* feat(MLU): Support PyTorch backend on MLU

  * MMCV support PyTorch backend on MLU

  * Add MLUDataParallel and MLUDistributedDataParallel

  * Add MLU operator support

* [Fix]: Fix PR comments and add IS_MLU to get device available check

* [Fix]: fix PR comments of dist_utils.py

* [Doc] Rewrite annotations of functions.

* [Docs] Rewrite annotation in distributed.py
Co-authored-by: Zaida Zhou <58739961+zhouzaida@users.noreply.github.com>

* [Docs] Fix lint
Co-authored-by: Zaida Zhou <58739961+zhouzaida@users.noreply.github.com>

* Speed up Registry initialization

This PR addresses https://github.com/open-mmlab/mmcv/issues/1843.

Instead of calling inspect.stack() to read the entire stack and its
associated source files from disk, walk up the stack to get only the
specific frame that we need (see [1] for additional information).

This makes imports in downstream projects ~2.5x faster in my local dev
environment. For mmaction2, for example:

Before:

    $ python -m timeit -n1 -r1 "from mmaction.apis import init_recognizer, inference_recognizer"
    1 loop, best of 1: 1.94 sec per loop

After:

    $ python -m timeit -n1 -r1 "from mmaction.apis import init_recognizer, inference_recognizer"
    1 loop, best of 1: 754 msec per loop

[1] https://stackoverflow.com/a/42636264/895769

* Add comment with PR tag

Explain why we avoid `inspect.stack()` with link to PR

* optimize forward

* fast backward

* fix bugs of grad input2

* [feat]:support find free port in _init_dist_slurm

* fix format

* Update mmcv/runner/dist_utils.py

should support port taken by a non-localhost address.
Co-authored-by: Mashiro <57566630+HAOCHENYE@users.noreply.github.com>

* Update dist_utils.py

Add Copyright.

* rename inner function

* Update mmcv/runner/dist_utils.py
Co-authored-by: Zaida Zhou <58739961+zhouzaida@users.noreply.github.com>

* fix dist_utils.py

change _is_port_in_use() criterion.

* Update dist_utils.py

rename _is_port_in_use to _is_free_port

* Update mmcv/runner/dist_utils.py
Co-authored-by: Zaida Zhou <58739961+zhouzaida@users.noreply.github.com>

* Update dist_utils.py

fix lint

* Update dist_utils.py

fix lint
Co-authored-by: Mashiro <57566630+HAOCHENYE@users.noreply.github.com>
Co-authored-by: Zaida Zhou <58739961+zhouzaida@users.noreply.github.com>

* [fix] fix collect_env() on Windows

* fix(utils): add back 'Cuda compilation tools' to nvcc info

* feat(utils): find default ccompiler

* fix(utils, test): parse cl.exe version instead of path

* fix: code style

* fix(tools): use distutils

* [fix] collect_env() returns GCC or MSVC

* [fix] unitest for collect_env()

* add comments

* test: add GCC to expected_keys

* fix: code style

* Update mmcv/utils/env.py

* Fix python3.10 ci

* fix build ci

* fix build ci

* fix onnx build in ci

* remove onnx optimizer

* remove python3.10 ci

* special treatment for py3.10

* fix build ci

* special treatment for python3.10 in requirements/test

* fix requirements/test

* update support for linearly learning rate decay

* Fix LinearAnnealingLrUpdaterHook, update LinearAnnealingMomentumUpdaterHook, add unit test

add docstring

add docstring

update linear lr momentum schedule test

fix ci

Fix CI

* add lr and momentum hook to runner and hooks package

add lr and momentum hook to runner and hooks package

* replace multi_optimziers with multi_optimizers
Co-authored-by: HAOCHENYE <21724054@zju.edu.cn>
Co-authored-by: Mashiro <57566630+HAOCHENYE@users.noreply.github.com>

* fix pt111 dist

* fix val step

* fix voxelization bug

* fix code

* rename grad_sampling_loc and grad_attn_weight

* recover cache initialize

* [Docs] Add more windows information in README

* minor fix

* rename MacOS to macOS

* implement corner pool with python for torch<1.5

* fix for torch130

* [fix] config ignore modules and functions

* add unitest

* add python3.10 ci

* fix github workflow

* [Fix] remove warnings in compilation

* [Fix] fix order of statements

* [Fix] fix build error on windows

* cannel buildextension wrapper

* Add .owners.yml to mark daily issue shift

* Update .owners.yml

fix end of file

* Add __copy__() to Config

* fix format in config.py

* fix format in test_copy

* fix format in config.py

* Update config.py

* Update tests/test_utils/test_config.py
Co-authored-by: Zaida Zhou <58739961+zhouzaida@users.noreply.github.com>
Co-authored-by: Zaida Zhou <58739961+zhouzaida@users.noreply.github.com>

* add nondeterministic voxelization op

* fix lint

* fix lint

* resolve comments

* fix lint

* add pytorch1.11 ci

* fix test parallel

* fix test parallel

* add worker_init_fn

* "Fix as comment"

* Fix format
Co-authored-by: Zaida Zhou <58739961+zhouzaida@users.noreply.github.com>

minor typo fix in `docs/en/understand_mmcv/data_process.md`

Fixed typo in `docs/en/understand_mmcv/cnn.md`. The `BaseModule` class implements `init_weights()` not `init_weight()`

* minor typo

minor typing bug. replaced `deprecated.py` to `deprecated_cfg.py`

* fixing the typo

as per the example, the directory name is `converters` not `converter`.

* Add .owners.yml to mark daily issue shift

* Update .owners.yml

fix end of file

* Fix bugs in  Voxelization op

* fix comments

* fix lint

* add comments

* Fix bugs in DynamicScatter op

* recover unittest

* add a comment as a reminder

* compatible to torch with lower version