All terms
Multimodal
Depth Estimation
Predicting how far each point in a scene is from the camera.
Definition
Depth estimation predicts the distance from the camera to each point in a scene. Single-camera methods guess depth from one image using patterns learned from past scenes, while two-camera or multi-view methods work it out from how the scene shifts between viewpoints. It is a key part of 3D understanding and feeds robot navigation, augmented reality, and self-driving car systems.