Do We Really Need Scene-specific Pose Encoders?

Yoli Shavit, Ron Ferens

Visual pose regression models estimate the camera pose from a query image with a single forward pass. Current models learn pose encoding from an image using deep convolutional netw…