Objects, attributes, actions, and poses are detected in individual camera views and projected into a 3D scene stitched from field-of-views of a network of cameras.
A garden scene. Estimated field-of-views of each camera are projected onto the scene map shown as colored contours. Mini-figures reflect the detected poses and facing directions.
A parking lot scene including a camera mounted on a moving vehicle. Estimated field-of-views of the moving camera are drawn on the scene map.
An auditorium scene including an IR camera.