Docket #: S16-237
3D Semantic Parsing of Indoor Space Point Clouds
Stanford researchers have developed a method that allows for 3D semantic parsing of indoor spaces. It receives a 3D point cloud input which is parsed into individual spaces and specific components, such as structural and furniture. This semantic output is provided in three differing granularities of resolution, formatted as points and 3D geometry.
Applications
- Architecture Engineering Construction and Facility Management - Automate processes in as-built Documentation and Building Information Modeling, providing insightful statistics.
- Virtual reality, Augmented reality, indoor navigation and robotic systems and computer vision - Can be used in any application that requires an understanding of the space of operation/interaction.
- Interior Design and Architecture - Provides an as-is model base for renovation projects and allows for effortless modifications to layout.
Advantages
- Able to handle large-scale spaces
- Hierarchical approach that can take advantage of recurrent layout configurations and patterns in man-made structures.
- Room parsing is parameter-free and doesn't focus on separating rooms by identifying wall surfaces
- Can process large point clouds with millions of points in linear time
- End-to-end automatic method without need for user input
Patents
- Issued: 10,424,065 (USA)
Similar Technologies
-
MEMS phased array for high-speed, random access variable focusing and control for LIDAR and 3D imaging S18-327MEMS phased array for high-speed, random access variable focusing and control for LIDAR and 3D imaging
-
Low-Power CMOS Image Sensor for Object Detection S13-419Low-Power CMOS Image Sensor for Object Detection
-
Resonant Scanning Design and Control for Fast Spatial Sampling S20-531Resonant Scanning Design and Control for Fast Spatial Sampling