How to do object recognition with Zed?

edit

I am working on a robot with Zed stereo camera. I could find zed warper on ROS and it outputs the camera left and right images, depth map, point cloud, odometry information. But I have no idea about how to combine these RGB images and point cloud information for object detection based on ROS. (I also find package like tabletop_objects to do point cloud recognition, but this will have nothing to do with RGB images). I wonder whether there are some common ways to combine these information. Thanks!