search

keras-team / keras-cv

Industry-strength Computer Vision workflows with Keras
Other
1.01k stars 330 forks source link

About YoloV8 License #2412

Closed heylamourding closed 7 months ago

heylamourding commented 7 months ago

Ultralytics YoloV8 is under AGPL-3.0 license. If I use keras-cv to train a YoloV8 model, can I use it commercially?

innat commented 7 months ago

I think we can use it commercially. The keras implementation is done by keras-team.

sachinprasadhs commented 7 months ago

To use Keras framework, there is no restriction.

Even from the Ultralytics licensing, it says it can be used for commercial purpose( look for permission table on top) https://github.com/ultralytics/ultralytics/blob/main/LICENSE

heylamourding commented 7 months ago

@sachinprasadhs thanks for reply. Yes. We can use ultralytics commercially but we have to disclose whole project due to AGPL license restriction.

heylamourding commented 7 months ago

@innat thanks for reply!

As keras-cv license is Apache 2.0, does mean that we can employ YOLOv8, trained via keras-cv, for commercial purposes without revealing our project details?

I notice a DISCLAIMER session in README.md. Is YoloV8 provide by third party and subject to separate license?

KerasCV provides access to pre-trained models via the keras_cv.models API. These pre-trained models are provided on an "as is" basis, without warranties or conditions of any kind. The following underlying models are provided by third parties, and are subject to separate licenses: StableDiffusion, Vision Transformer

innat commented 7 months ago

@heylamourding Keras implementaiton is done by keras-team (https://github.com/keras-team/keras-cv/pull/1711). So, if you use it, train it from scratch, it's fine.

About pretrained weight, here, the preset that ends with coco are pretrained weight. Now, this I'm not sure how these are generated (cc. @sachinprasadhs ).

Note, though the model implementation may be same but the performance gap exist in keras-cv yolov8 and official yolo-v8, due to lack of training strategies in keras-cv than official.

heylamourding commented 7 months ago

thanks @innat glad to hear that!