cvat-ai / cvat

Annotate better with CVAT, the industry-leading data engine for machine learning. Used and trusted by teams at any scale, for data of any scale.
https://cvat.ai
MIT License
12.58k stars 3k forks source link

Some practical small functions #8258

Open KTXKIKI opened 3 months ago

KTXKIKI commented 3 months ago

Actions before raising this issue

Is your feature request related to a problem? Please describe.

  1. Can CVAT support deduplication function for video and image data? This will be more efficient in some annotation aspects because in reality, when uploading videos or images, it is always necessary to filter and deduplicate, which will take up a lot of time

2.Currently, CVAT deletes images one by one. If batch deletion is possible, it would be more convenient

  1. Can the CVAT export function be selected through a switch to export datasets with annotations but without empty annotation images, or all images? Because sometimes some annotation tasks are not completed urgently and export is always done locally with scripts to remove empty annotation files, which is a bit annoying

4.Can CVAT support exporting JSON format with custom property Labelme

5.Perhaps a progress bar can be displayed under each project box to show how many tasks have been completed, or options can be added to view the overall project dashboard

Describe the solution you'd like

These are just some of my small suggestions. If there is any offense, don't pay attention I am working hard to learn front-end and back-end knowledge. These are competitive features that most platforms do not have. I will keep trying to develop them. If I develop them, I will submit a PR .

Salute to the great open-source workers! I hope CVAT becomes stronger and stronger!

Describe alternatives you've considered

No response

Additional context

No response

KTXKIKI commented 3 months ago

@Eldies Hello, do you have any suggestions for supporting labelme's JSON export

zhiltsov-max commented 3 months ago

Hi, thank you for sharing your ideas. I will comment just on the 1st point here - while I agree such a feature could be useful, there is a partial solution we use to filter the datasets currently. There is the near-duplicate removal (NDR) function in Datumaro, which you can use.

KTXKIKI commented 3 months ago

您好,感谢您分享您的想法。我在这里只评论第一点 - 虽然我同意这样的功能可能很有用,但我们目前使用部分解决方案来过滤数据集。Datumaro 中有近似重复删除 (NDR) 功能,您可以使用。

Thank you for your diligent reply. This method can also be used, but it is not an out of the box solution. However, please rest assured that I will continue to try and develop it