Yes, you’re not going to detect collision shapes on a Tilemap without adding any!
More important is that you don’t use a raycast to detect an intersection with the mouse in 2D physics like you would in 3D physics. You use the much faster/efficient Physics2D.OverlapPoint simply because this is 2D!
What if colliders are 2D but the camera is 3D (perspective)? In that case you have to use ScreenPointToRay and GetRaytIntersection. Do I understand this correctly?
Yes. As long as you remember that it’s 2D physics and it exists on the XY plane only. That query will sort by the Transform Z for you but the physics knows nothing about it. If all the colliders are on the same Z (which doesn’t affect physics) then the order is “undefined”.
I only mention this because I’ve encountered many devs who just grab 3D physics code and straight-up use it with 2D and don’t think about the fact that it’s not 3D.