Bebop 2 + Tensorflow

Hello,

I’m working on a ProofOfConcept application for my Bebop 2. I wish to integrate TFlite object detection to the H264VideoView in order to detect objects while driving the Drone.

The problem is that I’m getting wrong positions and values for the detected objects. After long research operations, I found out it’s related to the picture format. I think the video stream is not a flat classic pictures, it’s more like fish eye view, bubbled pictures.

Can you please guide me how to make the video stream work with the TF?
I will be happy if you have any suggestions.

One suggestion : use an Anafi drone instead of a Bebop :slight_smile:

This topic was automatically closed 2 days after the last reply. New replies are no longer allowed.