A system can estimate depth from one image, several views or depth-sensitive sensors. The resulting map represents relative or measured distance and can guide visual effects or scene understanding.
A single-image estimate can be uncertain or lack an accurate real-world scale. Occlusion, reflective surfaces and unfamiliar scenes can produce errors that affect downstream processing.
ELI5
Depth estimation tries to work out which parts of a scene are near or far. A flat picture does not directly provide every distance, so the system often has to infer them.
For example, an editing tool can estimate that a person is in front of a wall. That helps separate foreground from background, but hair edges or reflections can still confuse the estimate.
Does an image always provide exact distances?
No. Some approaches infer relative depth, and exact physical scale may need additional information.
Why does depth help editing?
It supplies spatial structure that can help distinguish foreground, background and object placement.
