Import annotations for video obect tracking

**Is your feature request related to a problem? Please describe.**
I do a lot of video object annotation and it is very time consuming and could maybe be improved.

**Describe the solution you'd like**
Maybe it is already possible and i just didn't find a way but i would know how i could predict the annotations with an object detection model and export them (and the videos) in label sutdio to check them and make some change if needed.
I use yolov5 to train my models and my annotation are in .txt, very different from the .json format in label studio for video object tracking.
Can you suggest a way to do that ? Maybe i can train my data in a different model that can predict and give me a .json format that i can import in labelstudio ?

I would be amazing if i can use my yolo trained model to predict directly the videos that i have on labelstudio.

Thank you !