autolabel objects

[BitStrawber]

Search before asking

• I have searched the YOLOv5 issues and found no similar feature requests.

Description

It's a large work to label the object one by one, especially when the datatset is pretty large. Maybe there could be something ued to generate label initially

Use case

No response

Additional

No response

Are you willing to submit a PR?

• Yes I'd like to help by submitting a PR!

[UltralyticsAssistant]

👋 Hello @BitStrawber, thank you for your interest in YOLOv5 🚀! Your suggestion regarding automated labeling is an interesting one, as managing large datasets can indeed be tedious.

For now, have you explored features like Ultralytics Label Studio integration, or tools like Roboflow for labeling assistance? These can potentially speed up the dataset preparation process. In addition — while YOLOv-possible solutions exploration here poweredRequest?) automating

[pderrenger]

@BitStrawber thank you for your suggestion! For automating initial labeling, you can consider tools like Roboflow Annotate, which supports features like pre-labeling using pre-trained models. Alternatively, you could apply a YOLOv5 model trained on a similar dataset to auto-generate labels and then refine them manually. For annotation tools and workflows, you can refer to the Train Custom Data Guide. Let us know if you have further questions!