open-mmlab/mmrotate

Roadmap of MMRotate

zytx121 opened this issue · 26 comments

zytx121 commented

We keep this issue open to collect feature requests from users and hear your voice. Our monthly release plan is also available here.

You can either:

Suggest a new feature by leaving a comment.
Vote for a feature request with 👍 or be against with 👎. (Remember that developers are busy and cannot respond to all feature requests, so vote for your most favorable one!)
Tell us that you would like to help implement one of the features in the list or review the PRs. (This is the greatest things to hear about!)

We also released our TODO list on the project page. Most of the TODO items are described in their corresponding issues (those labeled by Dev-RD) with detailed requirement documentation. Feel free to leave a message in the issue of any item and create a PR if you are interested in any of the item.

aldakata commented

I am down to help!

  • ❤️2
liuyanyi commented

Is there any plan for Rotated IoU Loss or Poly IoU Loss, it can easily improve detector performance. But both of them need cuda ops, i can help with python code, but I don't know much about cuda code. =.=

  • 👍4
yangxue0827 commented

@liuyanyi already in the plan @zytx121

HJoonKwon commented

Any plan for applying oriented bbox to instance segmentation?

  • 👍9
chandlerbing65nm commented

can you also make the mmrotate implementation of PIoU loss ?

  • 👍4
xxmomo commented

Will you add anchor-free model in furture? Maybe a few month later?

yangxue0827 commented

rotated_reppoints is an anchor free detector.

xxmomo commented
  • 👍2
yangxue0827 commented

Yes, we will. FCOS is on the way, done by @liuyanyi
You also can use the FCOS-OBB in OBBDetection instead.

xxmomo commented
aparico commented

@zytx121 @yangxue0827 Could you please add Darknet53 as backbone (from YOLOv4) and YOLO as detector?

  • 👍4
xxmomo commented
chandlerbing65nm commented

Pls add Masks in MMRotate

dimakan commented

Please support onnx/trt export.
directly or with mmdeploy (prefered)

  • 👍4
dimakan commented

Please support onnx/trt export. directly or with mmdeploy (prefered)

I noticed it stated as supported in the paper - really confusing:
https://www.arxiv-vanity.com/papers/2204.13317/
image

ziruoliu commented

Is there a plan to add 2022 CVPR “Oriented RepPoints for Aerial Object Detection”？

yangxue0827 commented

Already in processing (waiting to be merged), refer to #286

You can use it directly. @ziruoliu

kimlia545 commented

mmdet.datasets.pipelines.test_time_aug
When will "aug_test" be added?

hongxrsysu commented

(1) Draw the PR diagram in mmrotate!
(2) And the loss of visual val is also drawn, because we want to see whether it is over-fitting, and we can get a figure similar to this:
e8b76870f837437a3d34b69ff5cbac2

  • 👍2
ogencoglu commented

+1 for onnx conversion

CF-chen-feng-CF commented

onnx转换为+2

A1leilei commented

'OrientedRepPointsHead is already registered in models,情况该如何解决啊

timresink commented

Hi, I was wondering if there is any releases planned? There haven't been major updates for a while, and just curious if you guys are still actively developing

mohamedrekik commented

Hi, can we have yolov8 for Oriented Object Detection ?

yangxue0827 commented

Hi, can we have yolov8 for Oriented Object Detection ?

https://docs.ultralytics.com/zh/tasks/obb/#visual-samples

  • 👍1