search

microsoft / OmniParser

A simple screen parsing tool towards pure vision based GUI agent
Creative Commons Attribution 4.0 International
4.74k stars 357 forks source link

Provide another Object Detection model #9

Open masc-it opened 3 weeks ago

masc-it commented 3 weeks ago

Hey nice project,

Can you please, for the sake of true open source spirit, provide a non AGPL object detection model? YoloNAS, CenterNet, whatever you'd like.

Ultralytics' stuff is a pain to work with in industrial settings (we call it pizzo in italy) even if they've just iterated over the sacred Joseph Redmon's GPL yolov3.

Edit: I saw the LICENSE pull request. Until you have an AGPL components, the whole repo must be AGPL as well. It cannot be MIT, unless you've signed something special with them, which I dont think so.

Thanks.

yadong-lu commented 3 weeks ago

Thank you for your suggestion! We updated the model to huggingface and added the AGPL license: icon_detect/LICENSE · microsoft/OmniParser at main. Yes we are considering training alternatives to YOLO model which are under MIT license. In fact, we are working on OmniParser v2, which will be under MIT license as well. Stay tuned.