GitLab

* Upgrade the versions of pre-commit hooks

* update the versions of zh-cn.yaml

* init npu

* add npu extension and focal loss adapter

* clean code

* clean code

* clean code

* clean code

* fix autocast bugs on npu (#2273)

fix autocast bugs on npu (#2273)

* code format

* code format

* code format

* bug fix

* pytorch_npu_helper.hpp clean code

* Npu dev (#2306)

* fix autocast bugs on npu
* using scatter_kwargs in mmcv.device.scatter_gather

* raise ImportError when compile with npu

* add npu test case (#2307)

* add npu test case

* Update focal_loss.py

* add comment

* clean lint

* update dtype assert

* update DDP forward and comment

* fix bug
Co-authored-by: Zaida Zhou <58739961+zhouzaida@users.noreply.github.com>
Co-authored-by: ckirchhoff <515629648@qq.com>
Co-authored-by: Zaida Zhou <58739961+zhouzaida@users.noreply.github.com>

* support silu torch<1.7.0

* fix test

* fix test

* fix inplace

* fix inplace

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

* [Feature] Support RoipointPool3d with cambricon MLU backend

* [Feature] Support RoipointPool3d with cambricon MLU backend

* [Feature] Support RoipointPool3d with cambricon MLU backend

* [Feature] Support RoipointPool3d with cambricon MLU backend

* [Feature] Support RoipointPool3d with cambricon MLU backend

* [Feature] Support DeformRoiPool with cambricon MLU backend

* [Fix] Remove use of std library

* [Fix] Correct the error information

* [Refactor] Refactor test deform_roi_pool code

* [Fix] Fix judgment error

* [Fix] Modify the large tensor check
Co-authored-by: budefei <budefei@cambricon.com>

* [Feature] Support MaskedConv2d with cambricon MLU backend

* [Refactor] Refactor test masked_conv2d code
Co-authored-by: budefei <budefei@cambricon.com>

* [Feature] Support CARAFE with Cambricon MLU backend

* [Docs] Add comments for common funtions

* [Test] Add allclose test for carafe

* Remove print
Co-authored-by: zcyKTH <zcy19950525@gmail.com>
Co-authored-by: budefei <budefei@cambricon.com>

* support different border_mode in imrotate

* fix docstring

* small fix

* reorder the border_mode arg in imrotate

* fix lint

* polish docstring

* polish docstring

* Fix gradient cumulative optimizer when resuming

* add unit test

* add 'backend' for ColorJitter related functions

* add unittest

* fix unittest

* add bgr transpose and revise ut

* update unittest

* revise docstring

* [Feature] Support setting start epoch and interval epoch for PAVI Logger Hook

* [Feature] Update the coding style as the maintainer wish

* fix: default integer division or modulo by zero

* fix: runner.epoch is less than start and use self.get_epoch instead of runner.epoch

* feat: support for iter-based runner and fix the step bug

* feat: iter based hook

* feat: fix bug and coding style

* fix: coding style

* fix: coding style

* fix: graph may add in evaluation

* support prroipooling

* fix build bug

* fix bug of precision

* fix lint

* add ut

* add license and citation

* polish op and add type hint

* refactor ut

* add docstring for prroipool

* polish code

* format cu and cuh code

* fix format

* fix format

* fix typo
Co-authored-by: zhouzaida <zhouzaida@163.com>
Co-authored-by: Zaida Zhou <58739961+zhouzaida@users.noreply.github.com>

* add mps bbox overlap

* format

* update document and manifest

* update readme

* [Feature] Add support for MPS

* fix import error

* update ut

* fix error

* trigger CI

* use a unique basename for test file modules

* avoid bc-breaking

* [Feature] Support RoiPool with cambricon MLU backend

* [Docs] Update ops.md

* Add torch_meshgrid_ij wrapper due to PyTorch change

* Update torch_meshgrid name/doc/version implementation

* Make imports local

* add ut

* ignore ut when torch is not available
Co-authored-by: zhouzaida <zhouzaida@163.com>

* Fix batched_nms for rotated box.
Add type hint for nms.py

* Add test

* doc string

* revert symbolic hint

* fix max_coordinate

* add comment

* rename type

* fix typo docstring

* fix iou3d

* rename variables in test

* fix comments

* update

* fix lint

* fix lint

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

* Update chamfer_distance.py

* Update test_chamfer_distance.py

* fix

* Update chamfer_distance_cuda_kernel.cuh

* Update chamfer_distance_cuda_kernel.cuh

* fix

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

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

* [Feature] Support Psamask with cambricon MLU backend

* [Fix] refine the format according to lint check

* Replace std:max/min with conditional operators

* [Fix] Format head files

* [Feature] Support RoiAlignRotated with cambricon MLU backend

* [Fix] Remove std lib in mlu files

* [Fix] replace std::min/max with conditional operators

* [Fix] Fix the problem that the instance cannot reuse when overwriting the file client.

* Remove _overridden_backends and _overridden_prefixes, and pop the old instance when overwriting the file client.

* add unittest

* fix prefix error

* modify clear_backend to overridden_backend

* Delete redundant comments.

* fix is_module_wrapper

* test is_module_wrapper

* fix code style

* add iou3d

* revert deprecated python function

* fix lint

* replace 3d iou/nms calls for bev iou/nms

* skip test_rotated_feature_align.py for parrots
Signed-off-by: del-zhenwu <dele.zhenwu@gmail.com>

* fix lint
Co-authored-by: zhouzaida <zhouzaida@163.com>

* Fix the bug met in using nasfpn

Fix the bug met in using nasfpn which is mentioned at https://github.com/open-mmlab/mmdetection/issues/5987

.
Avoid the strong restrictions of _resize function in BaseMergeCell:
1. When Downsampling the feature map, the feature map's shape must be divisible by the target size. We pad zero around feature map before max_pool2d opt to make it always divisible. (line 102 ~ 107)
2. Considering the different downsampling scale of H and W, shape[-2] and shape[-1] are involed in the definition of kernel_size. (line 110)

* Update merge_cells.py

check flake8 & isort

* Update merge_cells.py

* Update merge_cells.py

yapf

* Update mmcv/ops/merge_cells.py

X_pad rename to padding_x
Co-authored-by: Mashiro <57566630+HAOCHENYE@users.noreply.github.com>

* Update merge_cells.py

format the code style after renameing the X_pad to padding_x

* Update test_merge_cells.py

Mainly test the downsampling resize in BaseMergeCell. The smaller target size is set to (14, 7), the classical feature map's size in the last few stages of the backbone, which will product different downsampling scales in different dims.

* Update test_merge_cells.py

add "# Copyright (c) OpenMMLab. All rights reserved."

* Update merge_cells.py

format the variable name

* Update test_merge_cells.py

Testing divisible and indivisible situations simultaneously

* Update mmcv/ops/merge_cells.py

fix the bug when h is indivisible and w is divisible, the pad_w will be padded unreasonable.
Co-authored-by: Zaida Zhou <58739961+zhouzaida@users.noreply.github.com>

* Update mmcv/ops/merge_cells.py

fix the bug when w is indivisible and h is divisible, the pad_h will be padded unreasonable.
Co-authored-by: Zaida Zhou <58739961+zhouzaida@users.noreply.github.com>

* fix undefined error

* Update merge_cells.py

make pad_h, pad_w more readable

* Update test_merge_cells.py

use @pytest.mark.parametrize instead of 'for' methor

* Update merge_cells.py

* Update test_merge_cells.py

isort

* Update merge_cells.py

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

* add pyupgrade

* add options for pyupgrade

* minor refinement

* [Impl] impl adjust hue

* Update docstring

* Fix docstring

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

* config dump fix
dump should not depend on how config was loaded only where it is dumped
changed checks for type to the given filename instead of the source filename

* defined test_fromdict and fixed dump function
it might be more convoluted than it needs to be but the formatting depends on the extension of the argument or the extension of the self.filename

* config dump defaults to returning pretty_text, test_fromdict renamed to test_dump_from_dict

* some reformatting in docstrings

* refine unittest

* fix unit test as comment

* Refine docstring

* minor refinement

* import mmcv

* fix lint
Co-authored-by: HAOCHENYE <21724054@zju.edu.cn>
Co-authored-by: Zaida Zhou <58739961+zhouzaida@users.noreply.github.com>

* [Enhance] Support register function.

* fix unittest error

* add docs and unittest of register function

* modify the docs

* fix version to 1.5.1

* Update docs/zh_cn/understand_mmcv/registry.md
Co-authored-by: Zaida Zhou <58739961+zhouzaida@users.noreply.github.com>

* Update docs/en/understand_mmcv/registry.md
Co-authored-by: Zaida Zhou <58739961+zhouzaida@users.noreply.github.com>

* refine the docs

* modify module_class to module
Co-authored-by: Zaida Zhou <58739961+zhouzaida@users.noreply.github.com>

* add rotated_feature_align cpu implementation

* add rotated_feature_align onnxruntime implementation

* Update code for advices from grimoire

Remove useless comment from mmcv/ops/csrc/pytorch/cpu/rotated_feature_align.cpp

Replace ambiguous function name atomicAdd in mmcv/ops/csrc/pytorch/cpu/rotated_feature_align.cpp

Simplify unit test with parameter in tests/test_ops/test_rotated_feature_align.py

Use fma in interpolate in mmcv/ops/csrc/onnxruntime/cpu/rotated_feature_align.cpp mmcv/ops/csrc/pytorch/cpu/rotated_feature_align.cpp

* Inline the function to reduce the overhead of the function call

Use fma in interpolate

* ensure type of filename is str

* check filename for func: fromfile

* add ut for fromfile