How trasform image points in map frame

[Georgegal00]

The goal of my project is to provide a reference_path for the vehicle to follow the lane. It is done using a lane detection network that returns points to me in 2D image coordinates. The goal is to transfrom these points into the frame map to get a valid path. I would like to understand how to transform these points in the right way. I'm using the carla_autoware_interface, so the simulation is in Carla.From the topic camera_info I see that the camera frame is traffic_light_left_camera/camera_optical_link, and with : `ros2 run tf2_ros tf2_echo map traffic_light_left_camera/camera_optical_link
[INFO] [1739626472.451398971] [tf2_echo]: Waiting for transform map -> traffic_light_left_camera/camera_optical_link: Invalid frame ID “map” passed to canTransform argument target_frame - frame does not exist
At time 472.850007046

• Translation: [-485.596, -74.916, 1.968]
• Rotation: in Quaternion [-0.702, 0.121, -0.121, 0.691]
• Rotation: in RPY (radian) [-1.586, -0.002, -0.344]
• Rotation: in RPY (degree) [-90.857, -0.088, -19.725]
• Matrix:
  0.941 -0.004 0.337 -485.596
  -0.337 -0.015 0.941 -74.916
  0.002 -1.000 -0.015 1.968
  0.000 0.000 0.000 1.000`. Also from the camera_info topic are available the matrices needed to perform perspective transfrom, I would like to know if instead the spawn_point configuration in object.json is related to the world or the vehicle. To transfrom image coordinates to frame map what steps should I follow? Should I transfrom points to 3D camera coordinates first and then apply the transformation? And as for the transfrom is it okay to use the one shown above? Or should I go backwards in the tree, then go through sensor_base_kit -> base_link -> map?Any advice will be extremely helpful, thank you.