Skip to content

Releases: open-mmlab/mmrotate

v0.3.4

01 Feb 11:59
@RangiLyu RangiLyu
v0.3.4
7755aa5
This commit was created on GitHub.com and signed with GitHub’s verified signature. The key has expired.
GPG key ID: 4AEE18F83AFDEB23
Expired
Learn about vigilant mode.
Compare
Choose a tag to compare
v0.3.4 Latest
Latest

Improvements

  • Use iof for RRandomCrop validation (#660)
  • Upgrade e2cnn version (#713)
  • Support empty patch in Rotate Transform (#712)

Bug Fixes

  • Fix scikit-learn installation (#658)
  • Fix deprecated np.bool (#685)
  • Fix using multiprocessing when formatting (#679)

Documentations

  • Minor correction in the documentation (#643)

Contributors

A total of 3 developers contributed to this release.
Thanks @nijkah, @jistiak, @RangiLyu

New Contributors

Full Changelog: v0.3.3...v0.3.4

Contributors

RangiLyu, jistiak, and nijkah
Assets 2
Loading
0 Join discussion

MMRotate v1.0.0rc1 Release

03 Jan 09:00
@zytx121 zytx121
v1.0.0rc1
5d0491c
This commit was created on GitHub.com and signed with GitHub’s verified signature. The key has expired.
GPG key ID: 4AEE18F83AFDEB23
Expired
Learn about vigilant mode.
Compare
Choose a tag to compare
MMRotate v1.0.0rc1 Release Pre-release
Pre-release

Highlights

  • Support RTMDet rotated object detection models. The technical report of RTMDet is on arxiv (#662)
  • Support H2RBox. (#644)

New Features

  • Support PSC (#617)
  • Add projects/ folder and give an example for communities to contribute their projects. (#627)
  • Support DIOR Dataset. (#639)

Bug Fixes

  • Fix get_flops.py in 1.x. (#646)
  • Fix Windows CI. (#621)
  • Fix error in rbbox_overlaps. (#620)

Improvements

  • Deprecating old type alias due to new version of numpy (#674)
  • Use iof in RRandomCrop. (#660)
  • Modify keys in dataset.metainfo to lower case. (#654)
  • Add torch 1.13 in CI. (#661)
  • Add dockerfile in 1.x. (#631)
  • Use mmengine in torchserve deployment. (#616)
  • Add .pre-commit-config-zh-cn.yaml. (#630)

Contributors

A total of 11 developers contributed to this release.

Thanks @yxzhao2022 @yuyi1005 @YanxingLiu @nijkah @RangeKing @austinmw @liuyanyi @yangxue0827 @zytx121 @RangiLyu @ZwwWayne

New Contributors

Full Changelog: v1.0.0rc0...v1.0.0rc1

Contributors

yuyi1005, zytx121, and 9 other contributors
Assets 2
Loading
0 Join discussion

MMRotate v1.0.0rc0 Release

07 Nov 14:05
@zytx121 zytx121
v1.0.0rc0
766185e
Compare
Choose a tag to compare
MMRotate v1.0.0rc0 Release Pre-release
Pre-release

We are excited to announce the release of MMRotate 1.0.0rc0.
MMRotate 1.0.0rc0 is the first version of MMRotate 1.x, a part of the OpenMMLab 2.0 projects.
Built upon the new training engine, MMRotate 1.x unifies the interfaces of dataset, models, evaluation, and visualization with faster training and testing speed.

Highlights

  1. New engines. MMRotate 1.x is based on MMEngine, which provides a general and powerful runner that allows more flexible customizations and significantly simplifies the entrypoints of high-level interfaces.

  2. Unified interfaces. As a part of the OpenMMLab 2.0 projects, MMRotate 1.x unifies and refactors the interfaces and internal logics of train, testing, datasets, models, evaluation, and visualization. All the OpenMMLab 2.0 projects share the same design in those interfaces and logics to allow the emergence of multi-task/modality algorithms.

  3. New BoxType design. We support data structures RotatedBoxes and QuadriBoxes to encapsulate different kinds of bounding boxes. We are migrating to use data structures of boxes to replace the use of pure tensor boxes. This will unify the usages of different kinds of bounding boxes in MMDetection 3.x and MMRotate 1.x to simplify the implementation and reduce redundant codes.

  4. Stronger visualization. We provide a series of useful tools which are mostly based on brand-new visualizers. As a result, it is more convenient for the users to explore the models and datasets now.

Breaking Changes

We briefly list the major breaking changes here.
We will update the migration guide to provide complete details and migration instructions.

Dependencies

  • MMRotate 1.x relies on MMEngine to run. MMEngine is a new foundational library for training deep learning models in OpenMMLab 2.0 models. The dependencies of file IO and training are migrated from MMCV 1.x to MMEngine.
  • MMRotate 1.x relies on MMCV>=2.0.0rc2. Although MMCV no longer maintains the training functionalities since 2.0.0rc0, MMRotate 1.x relies on the data transforms, CUDA operators, and image processing interfaces in MMCV. Note that the package mmcv is the version that provide pre-built CUDA operators and mmcv-lite does not since MMCV 2.0.0rc0, while mmcv-full has been deprecated.
  • MMRotate 1.x relies on MMDetection>=3.0.0rc2.

Training and testing

  • MMRotate 1.x uses Runner in MMEngine rather than that in MMCV. The new Runner implements and unifies the building logic of dataset, model, evaluation, and visualizer. Therefore, MMRotate 1.x no longer maintains the building logics of those modules in mmrotate.train.apis and tools/train.py. Those code have been migrated into MMEngine. Please refer to the migration guide of Runner in MMEngine for more details.
  • The Runner in MMEngine also supports testing and validation. The testing scripts are also simplified, which has similar logic as that in training scripts to build the runner.
  • The execution points of hooks in the new Runner have been enriched to allow more flexible customization. Please refer to the migration guide of Hook in MMEngine for more details.
  • Learning rate and momentum scheduling has been migrated from Hook to Parameter Scheduler in MMEngine. Please refer to the migration guide of Parameter Scheduler in MMEngine for more details.

Configs

  • The Runner in MMEngine uses a different config structures to ease the understanding of the components in runner. Users can refer to the migration guide in MMEngine for migration details.
  • The file names of configs and models are also refactored to follow the new rules unified across OpenMMLab 2.0 projects.

Dataset

The Dataset classes implemented in MMRotate 1.x all inherits from the BaseDataset in MMEngine.

  • All the datasets support to serialize the data list to reduce the memory when multiple workers are built to accelerate data loading.

Data Transforms

The data transforms in MMRotate 1.x all inherits from those in MMCV>=2.0.0rc2, which follows a new convention in OpenMMLab 2.0 projects.
The changes are listed as below:

  • The interfaces are also changed. Please refer to the API Reference
  • The functionality of some data transforms (e.g., Rotate) are decomposed into several transforms.

Model

The models in MMRotate 1.x all inherits from BaseModel in MMEngine, which defines a new convention of models in OpenMMLab 2.0 projects. Users can refer to the tutorial of model in MMengine for more details. Accordingly, there are several changes as the following:

  • The model interfaces, including the input and output formats, are significantly simplified and unified following the new convention in MMRotate 1.x. Specifically, all the input data in training and testing are packed into inputs and data_samples, where inputs contains model inputs like a list of image tensors, and data_samples contains other information of the current data sample such as ground truths and model predictions. In this way, different tasks in MMRotate 1.x can share the same input arguments, which makes the models more general and suitable for multi-task learning.
  • The model has a data preprocessor module, which is used to pre-process the input data of model. In MMRotate 1.x, the data preprocessor usually does necessary steps to form the input images into a batch, such as padding. It can also serve as a place for some special data augmentations or more efficient data transformations like normalization.
  • The internal logic of model have been changed. In MMRotate 0.x, model used forward_train and simple_test to deal with different model forward logics. In MMRotate 1.x and OpenMMLab 2.0, the forward function has three modes: loss, predict, and tensor for training, inference, and tracing or other purposes, respectively. The forward function calls self.loss(), self.predict(), and self._forward() given the modes loss, predict, and tensor, respectively.

Evaluation

MMRotate 1.x mainly implements corresponding metrics for each task, which are manipulated by Evaluator to complete the evaluation.
In addition, users can build evaluator in MMRotate 1.x to conduct offline evaluation, i.e., evaluate predictions that may not produced by MMRotate, prediction follows our dataset conventions. More details can be find in the Evaluation Tutorial in MMEngine.

Visualization

The functions of visualization in MMRotate 1.x are removed. Instead, in OpenMMLab 2.0 projects, we use Visualizer to visualize data. MMRotate 1.x implements RotLocalVisualizer to allow visualization of ground truths, model predictions, and feature maps, etc., at any place. It also supports to dump the visualization data to any external visualization backends such as Tensorboard and Wandb.

Improvements

  • Support quadrilateral box detection (#520)
  • Support RotatedCocoMetric (#557)
  • Support COCO style annotations (#582)
  • Support two new SAR datasets: RSDD and SRSDD (#591)

Ongoing changes

  1. Test-time augmentation: is not implemented yet in this version due to limited time slot. We will support it in the following releases with a new and simplified design.
  2. Inference interfaces: a unified inference interfaces will be supported in the future to ease the use of released models.
  3. Interfaces of useful tools that can be used in notebook: more useful tools that implemented in the tools/ directory will have their python interfaces so that they can be used through notebook and in downstream libraries.
  4. Documentation: we will add more design docs, tutorials, and migration guidance so that the community can deep dive into our new design, participate the future development, and smoothly migrate downstream libraries to MMRotate 1.x.

Contributors

A total of 8 developers contributed to this release.
Thanks @DonggeunYu @k-papadakis @liuyanyi @yangxue0827 @jbwang1997 @zytx121 @RangiLyu @ZwwWayne

New Contributors

Full Changelog: v0.3.3...v1.0.0rc0

Contributors

zytx121, liuyanyi, and 6 other contributors
Assets 2
Loading

v0.3.3

27 Oct 08:46
@RangiLyu RangiLyu
v0.3.3
4bbef14
Compare
Choose a tag to compare
v0.3.3

Bug Fixes

  • Fix reppoint bug fix when negative image training (#396)
  • Fix bug in oriented_reppoints_head.py (#424)
  • Fix mmcv-full version (#423)

Improvements

  • Update issue templates to main branch (#579)
  • Fix lint of dev branch (#578)

Documentations

  • Update citation (#425)
  • Fix markdown version when building docs (#414)

Contributors

A total of 5 developers contributed to this release.
Thanks @yangxue0827, @ZwwWayne, @MinkiSong, @zytx121, @RangiLyu

Contributors

zytx121, RangiLyu, and 3 other contributors
Assets 2
Loading

v0.3.2

06 Jul 14:10
@zytx121 zytx121
v0.3.2
c62f148
This commit was created on GitHub.com and signed with GitHub’s verified signature. The key has expired.
GPG key ID: 4AEE18F83AFDEB23
Expired
Learn about vigilant mode.
Compare
Choose a tag to compare
v0.3.2

Changelog

v0.3.2 (6/7/2022)

Highlight

  • Support Oriented Reppoints (CVPR'22) (#286)
  • Support ConvNeXt backbone (CVPR'22) (#343)

New Features

  • Support RMosaic. (#344)

Bug Fixes

  • Fix max_coordinate in multiclass_nms_rotated. (#346)
  • Fix bug in PolyRandomRotate. (#366)
  • Fix memory shortage when using huge_image_demo.py (#368)

Improvements

  • Update README.md and INSTALL.md. (#342)
  • Fix typo in rotated_fcos_head. (#354)
  • Update checkpoint and eval interval of base config. (#347)
  • Fix mdformat version to support python3.6 & Add mim to extras_require in setup.py (#359)
  • Add mim test in CI (#374)

Contributors

A total of 9 developers contributed to this release.
Thanks @LiWentomng @heiyuxiaokai @JinYuannn @sltlls @liuyanyi @yangxue0827 @jbwang1997 @zytx121 @ZwwWayne

New Contributors

Full Changelog: v0.3.1...v0.3.2

Contributors

zytx121, liuyanyi, and 7 other contributors
Assets 2
Loading

v0.3.1

06 Jun 14:44
@zytx121 zytx121
v0.3.1
6eb7a27
This commit was created on GitHub.com and signed with GitHub’s verified signature. The key has expired.
GPG key ID: 4AEE18F83AFDEB23
Expired
Learn about vigilant mode.
Compare
Choose a tag to compare
v0.3.1

Highlight

New Features

  • Update PolyRandomRotate to support discrete angle value. (#281)
  • Support RRandomCrop. (#322)
  • Support mask in merge_results and huge_image_demo.py. (#280)
  • Support don't filter images without ground truths. (#323)
  • Add MultiImageMixDataset in build_dataset. (#331)

Bug Fixes

  • Fix error in Windows CI. (#324)
  • Fix data path error in config files. (#328)
  • Fix bug when visualize the HRSC2016 detect results. (#329)

Improvements

  • Add torchserve doc in zh_cn. (#287)
  • Fix doc typo in README. (#284)
  • Configure Myst-parser to parse anchor tag (#305 #308)
  • Replace markdownlint with mdformat for avoiding installing ruby. (#306)
  • Fix typo about split gap of multi scale. (#272)

Contributors

A total of 7 developers contributed to this release.
Thanks @liuyanyi @nijkah @remi-or @yangxue0827 @jbwang1997 @zytx121 @ZwwWayne

New Contributors

Full Changelog: v0.3.0...v0.3.1

Contributors

zytx121, liuyanyi, and 5 other contributors
Assets 2
Loading

v0.3.0

30 Apr 09:11
@ZwwWayne ZwwWayne
v0.3.0
d80310a
This commit was created on GitHub.com and signed with GitHub’s verified signature. The key has expired.
GPG key ID: 4AEE18F83AFDEB23
Expired
Learn about vigilant mode.
Compare
Choose a tag to compare
v0.3.0

Highlight

  • Support TorchServe (#160)
  • Support Rotated ATSS (CVPR'20) (#179)

New Features

  • Update performance of ReDet on HRSC2016. (#203)

  • Upgrage visualization to custom colors of different classes. This requires mmdet>=2.22.0. (#187, #267, #270)

  • Update Stable KLD, which solve the Nan issue of KLD training. (#183)

  • Support setting dataloader arguments in config and add functions to handle config compatibility. (#215)
    The comparison between the old and new usages is as below.

    Before v0.2.0 Since v0.3.0 
    data = dict(
    samples_per_gpu=2, workers_per_gpu=2,
    train=dict(type='xxx', ...),
    val=dict(type='xxx', samples_per_gpu=4, ...),
    test=dict(type='xxx', ...),
)
    		 
    # A recommended config that is clear
data = dict(
    train=dict(type='xxx', ...),
    val=dict(type='xxx', ...),
    test=dict(type='xxx', ...),
    # Use different batch size during inference.
    train_dataloader=dict(samples_per_gpu=2, workers_per_gpu=2),
    val_dataloader=dict(samples_per_gpu=4, workers_per_gpu=4),
    test_dataloader=dict(samples_per_gpu=4, workers_per_gpu=4),
)

# Old style still works but allows to set more arguments about data loaders
data = dict(
    samples_per_gpu=2,  # only works for train_dataloader
    workers_per_gpu=2,  # only works for train_dataloader
    train=dict(type='xxx', ...),
    val=dict(type='xxx', ...),
    test=dict(type='xxx', ...),
    # Use different batch size during inference.
    val_dataloader=dict(samples_per_gpu=4, workers_per_gpu=4),
    test_dataloader=dict(samples_per_gpu=4, workers_per_gpu=4),
)

  • Add get_flops tool (#176)

Bug Fixes

  • Fix bug about rotated anchor inside flags. (#197)
  • Fix Nan issue of GWD. (#206)
  • Fix bug in eval_rbbox_map when labels_ignore is None. (#209)
  • Fix bug of 'RoIAlignRotated' object has no attribute 'output_size' (#213)
  • Fix bug in unit test for datasets. (#222)
  • Fix bug in rotated_reppoints_head. (#246)
  • Fix GPG key error in CI and docker. (#269)

Improvements

  • Update citation of mmrotate in README.md (#263)
  • Update the introduction of SASM (AAAI'22) (#184)
  • Fix doc typo in Config File and Model Zoo. (#199)
  • Unified RBox definition in doc. (#234)

Contributors

A total of 8 developers contributed to this release.
Thanks @nijkah @GamblerZSY @liuyanyi @yangxue0827 @grimoire @jbwang1997 @zytx121 @ZwwWayne

New Contributors

Full Changelog: v0.2.0...v0.3.0

Contributors

grimoire, zytx121, and 6 other contributors
Assets 2
Loading

MMRotate V0.2.0 Release

01 Apr 14:49
@ZwwWayne ZwwWayne
v0.2.0
b78bab2
This commit was created on GitHub.com and signed with GitHub’s verified signature. The key has expired.
GPG key ID: 4AEE18F83AFDEB23
Expired
Learn about vigilant mode.
Compare
Choose a tag to compare
MMRotate V0.2.0 Release

New Features

Bug Fixes

  • Remove in-place operations in rbbox_overlaps (#155)
  • Fix bug in docstring. (#137)
  • Fix bug in HRSCDataset with clasesswise=ture (#175)

Improvements

  • Add Chinese translation of docs/zh_cn/tutorials/customize_dataset.md (#65)
  • Add different seeds to different ranks (#102)
  • Update from-scratch install script in install.md (#166)
  • Improve the arguments of all mmrotate scripts (#168)

Contributors

A total of 6 developers contributed to this release.
Thanks @zytx121 @yangxue0827 @ZwwWayne @jbwang1997 @canoe-Z @matrixgame2018

New Contributors

Full Changelog: v0.1.1...v0.2.0

Contributors

canoe-Z, zytx121, and 4 other contributors
Assets 2
Loading

MMRotate V0.1.1 Release

14 Mar 05:41
@ZwwWayne ZwwWayne
v0.1.1
c5bf348
This commit was created on GitHub.com and signed with GitHub’s verified signature. The key has expired.
GPG key ID: 4AEE18F83AFDEB23
Expired
Learn about vigilant mode.
Compare
Choose a tag to compare
MMRotate V0.1.1 Release

New Features

Bug Fixes

  • Fix URL error of Swin pretrained model (#111)
  • Fix bug for SASM during training (#105)
  • Fix rbbox_overlaps abnormal when the box is too small (#61)
  • Fix bug for visualization (#12, #81)
  • Fix stuck when compute mAP (#14, #52)
  • Fix 'RoIAlignRotated' object has no attribute 'out_size' bug (#51)
  • Add missing init_cfg in dense head (#37)
  • Fix install an additional mmcv (#17)
  • Fix typos in docs (#3, #11, #36)

Improvements

  • Move eval_rbbox_map from mmrotate.datasets to mmrotate.core.evaluation (#73)
  • Add Windows CI (#31)
  • Add copyright commit hook (#30)
  • Add Chinese translation of docs/zh_cn/get_started.md (#16)
  • Add Chinese translation of docs/zh_cn/tutorials/customize_runtime.md (#22)
  • Add Chinese translation of docs/zh_cn/tutorials/customize_config.md (#23)
  • Add Chinese translation of docs/zh_cn/tutorials/customize_models.md (#27)
  • Add Chinese translation of docs/zh_cn/model_zoo.md (#28)
  • Add Chinese translation of docs/zh_cn/faq.md (#33)

Contributors

A total of 13 developers contributed to this release.
Thanks @zytx121 @yangxue0827 @jbwang1997 @liuyanyi @DangChuong-DC @RangeKing @liufeinuaa @np-csu @akmalulkhairin @SheffieldCao @BrotherHappy @Abyssaledge @q3394101

New Contributors

Full Changelog: v0.1.0...v0.1.1

Contributors

np-csu, zytx121, and 14 other contributors
Assets 2
Loading

MMRotate V0.1.0 Release

18 Feb 09:42
@ZwwWayne ZwwWayne
v0.1.0
6519a36
This commit was created on GitHub.com and signed with GitHub’s verified signature. The key has expired.
GPG key ID: 4AEE18F83AFDEB23
Expired
Learn about vigilant mode.
Compare
Choose a tag to compare
MMRotate V0.1.0 Release

Initial Release of MMRotate, support the following algorithms

Assets 2
Loading