Abstract: Disclosed are techniques for generating and employing computer vision models of a structural environment. Real video is received of physical actions being performed in a physical space within the structural environment. A virtual replica of the physical space and objects within the physical space are generated, based on environment data corresponding to the physical space. Synthetic video that represents virtual actions being performed is generated using the virtual replica. A computer vision model of the structural environment is trained, based on the real video and the synthetic video. A real video stream of subsequent actions being performed in the physical space is received. Perception metadata that represents the subsequent actions being performed is generated, by providing the real video stream to a perception pipeline that uses the computer vision model of the structural environment. The perception metadata is aggregated, and a corresponding visualization is generated.

Abstract: Disclosed are techniques for generating and employing computer vision models of a structural environment. Real video is received of physical actions being performed in a physical space within the structural environment. A virtual replica of the physical space and objects within the physical space are generated, based on environment data corresponding to the physical space. Synthetic video that represents virtual actions being performed is generated using the virtual replica. A computer vision model of the structural environment is trained, based on the real video and the synthetic video. A real video stream of subsequent actions being performed in the physical space is received. Perception metadata that represents the subsequent actions being performed is generated, by providing the real video stream to a perception pipeline that uses the computer vision model of the structural environment. The perception metadata is aggregated, and a corresponding visualization is generated.