Clarify licensing

[savchenko]

Ref. https://old.reddit.com/r/computervision/comments/1e3uxro/ultralytics_new_agpl30_license_exploiting/

According to Ultralytics, both the training code and the models produced by that code are covered by AGPL-3.0. This means if you use their framework to train a model, that model and your software application that uses the model must also be open-sourced under the same license.

Could you please confirm that your YOLO implementation is indeed MIT-licensed?

Thank you.

[Eric-Canas]

Hi!

I'm not really an expert on licensing. So my unique answer is I don't really know. I'll do further research about it. If you are pretty sure the repo shoul be AGPL-3.0 licensed I'll change it.

I did it MIT, as I wanted to Open Source it without further requirements. Training code is not in the library as it is only a yolo.train() line from Ultralytics library, and models are open sourced in the release page.

The library just download the public model and runs it. You think license for the whole repo should be changed then? Or I should just put that license on the model's release?

I'm not an expert in the matter, so if you are confident on telling that the whole repo license should be changed, I'll do it.

Thanks!

[savchenko]

Let me preface it by saying that I am not a lawyer and this is not a legal advice.

In my personal opinion, if repository does not include AGPL content, but allows (at user's discretion) to download it from elsewhere, then there is no issue.

To be 100% safe, I would add a single-line prompt and/or a notice to README along the lines of This will download AGPL-3.0 model, do you agree? [Y/N].

[martinkielhorn-hexagon]

If your models, such as qrdet-n.pt, are derived from Ultralytics' weights and/or trained using their code, they would likely fall under the AGPL-3.0 license, which imposes specific requirements, particularly for commercial use. While you are generously providing the code for free and are under no obligation to meet individual requests, ensuring clarity around licensing is essential for users to make informed decisions.

If your work is based on Ultralytics' models or training code, it would be helpful to clearly reflect this in your repository. For example, adding a note in the README or a prompt during downloads to inform users about the AGPL-3.0 implications would greatly enhance transparency.

Alternatively, you might consider using models like YOLOv3 from MMDetection, which are licensed under Apache 2.0 and may provide more flexibility for commercial projects.

Thank you for your contributions to the open-source community!