Collision Detection/Avoidance Integration with QGroundControl

Hello, I’m really new to autonomous drone programming. If I remember correctly, to build an autonomous drone you need a waypoint mission planner software and an object detection (birds, other drones etc.) and collision avoidance program. How am I supposed to integrate these two together so that if something is detected the drone will avoid it but still maintain the path specified on QGroundControl? I come from a hard core Keras with Tensorflow backend background.

Please go through the obstacle avoidance documentation:

You could have a look at this post as well, it will helps