Systems and methods for 2D to 3D conversion

1. A system comprising:

one or more memory units; and

one or more computer processors communicatively coupled to the one or more memory units and configured to:

access a video generated by a first physical camera located within a physical environment;

identify, by analyzing a video frame of the video, an object of interest in the video frame, the object of interest corresponding to a physical object that is physically located within the physical environment;

display, in a graphical user interface, a virtual three-dimensional (3D) environment that corresponds to the physical environment, the virtual 3D environment comprising:

a first virtual camera that corresponds to the first physical camera; and

a second virtual camera that corresponds to a second physical camera located within the physical environment;

access, from a calibrations database, a plurality of settings of the first physical camera and a plurality of settings of the second physical camera;

configure a plurality of settings of the first virtual camera to match the plurality of settings of the first physical camera such that a field of view of the first virtual camera in the virtual 3D environment is identical to a field of view of the first physical camera in the physical environment;

configure a plurality of settings of the second virtual camera to match the plurality of settings of the second physical camera such that a field of view of the second virtual camera in the virtual 3D environment is identical to a field of view of the second physical camera in the physical environment; and

project the identified object of interest into the virtual 3D environment using the configured first and second virtual cameras.

2. The system of claim 1 , the one or more memory units are further configured to:

create a first synthetic depth map for the first virtual camera, wherein the first synthetic depth map provides a one-to-one mapping of each 2D pixel created by the first physical camera and a corresponding space of the first virtual camera in the virtual 3D environment; and

create a second synthetic depth map for the second virtual camera, wherein the second synthetic depth map provides a one-to-one mapping of each 2D pixel created by the second physical camera and a corresponding space of the second virtual camera in the virtual 3D environment;

create a camera matrix comprising the first and second synthetic depth maps.

3. The system of claim 2 , wherein projecting the identified object of interest into the virtual 3D environment comprises:

determining a 2D coordinate of the identified object of interest; and

converting the 2D coordinate to a location within the virtual 3D environment using the camera matrix.

4. The system of claim 2 , wherein the camera matrix enables the first and second physical cameras to be aware of detection capabilities of other physical cameras using a shared unified coordinate system.

5. The system of claim 1 , wherein identifying the object of interest comprises utilizing a convolution neural network architecture.

6. The system of claim 1 , wherein:

the first virtual camera is placed in the virtual 3D environment at a same latitude, longitude, and altitude as the first physical camera is positioned in the physical environment; and

the second virtual camera is placed in the virtual 3D environment at a same latitude, longitude, and altitude as the second physical camera is positioned in the physical environment.

7. The system of claim 1 , wherein the settings of the first and second virtual cameras and the first and second physical cameras each comprise:

a surge setting;

a sway setting;

a heave setting;

a roll setting;

a pitch setting; and

a yaw setting.

8. The system of claim 7 , wherein the settings of the first and second virtual cameras and the first and second physical cameras further comprise:

a radial distortion; and

a tangential distortion.

9. A method by a computing system, the method comprising:

accessing a video generated by a first physical camera located within a physical environment;

identifying, by analyzing a video frame of the video, an object of interest in the video frame, the object of interest corresponding to a physical object that is physically located within the physical environment;

displaying, in a graphical user interface, a virtual three-dimensional (3D) environment that corresponds to the physical environment, the virtual 3D environment comprising:

a first virtual camera that corresponds to the first physical camera; and

a second virtual camera that corresponds to a second physical camera located within the physical environment;

accessing, from a calibrations database, a plurality of settings of the first physical camera and a plurality of settings of the second physical camera;

configuring a plurality of settings of the first virtual camera to match the plurality of settings of the first physical camera such that a field of view of the first virtual camera in the virtual 3D environment corresponds to a field of view of the first physical camera in the physical environment;

configuring a plurality of settings of the second virtual camera to match the plurality of settings of the second physical camera such that a field of view of the second virtual camera in the virtual 3D environment corresponds to a field of view of the second physical camera in the physical environment; and

projecting the identified object of interest into the virtual 3D environment using the configured first and second virtual cameras.

10. The method of claim 9 , further comprising:

creating a first synthetic depth map for the first virtual camera, wherein the first synthetic depth map provides a one-to-one mapping of each 2D pixel created by the first physical camera and a corresponding space of the first virtual camera in the virtual 3D environment;

creating a second synthetic depth map for the second virtual camera, wherein the second synthetic depth map provides a one-to-one mapping of each 2D pixel created by the second physical camera and a corresponding space of the second virtual camera in the virtual 3D environment; and

creating a camera matrix comprising the first and second synthetic depth maps.

11. The method of claim 10 , wherein projecting the identified object of interest into the virtual 3D environment comprises:

determining a 2D coordinate of the identified object of interest; and

converting the 2D coordinate to a location within the virtual 3D environment using the camera matrix.

12. The method of claim 9 , wherein identifying the object of interest comprises utilizing a convolution neural network architecture.

13. The method of claim 9 , wherein:

the first virtual camera is placed in the virtual 3D environment at a same latitude, longitude, and altitude as the first physical camera is positioned in the physical environment; and

the second virtual camera is placed in the virtual 3D environment at a same latitude, longitude, and altitude as the second physical camera is positioned in the physical environment.

14. The method of claim 9 , wherein the settings of the first and second virtual cameras and the first and second physical cameras each comprise:

a surge setting;

a sway setting;

a heave setting;

a roll setting;

a pitch setting; and

a yaw setting.

15. The method of claim 14 , wherein the settings of the first and second virtual cameras and the first and second physical cameras further comprise:

a radial distortion; and

a tangential distortion.

16. One or more computer-readable non-transitory storage media embodying instructions that, when executed by a processor, cause the processor to perform operations comprising:

accessing a video generated by a first physical camera located within a physical environment;

identifying, by analyzing a video frame of the video, an object of interest in the video frame, the object of interest corresponding to a physical object that is physically located within the physical environment;

displaying, in a graphical user interface, a virtual three-dimensional (3D) environment that corresponds to the physical environment, the virtual 3D environment comprising:

a first virtual camera that corresponds to the first physical camera; and

a second virtual camera that corresponds to a second physical camera located within the physical environment;

accessing, from a calibrations database, a plurality of settings of the first physical camera and a plurality of settings of the second physical camera;

configuring a plurality of settings of the first virtual camera to match the plurality of settings of the first physical camera such that a field of view of the first virtual camera in the virtual 3D environment corresponds to a field of view of the first physical camera in the physical environment;

configuring a plurality of settings of the second virtual camera to match the plurality of settings of the second physical camera such that a field of view of the second virtual camera in the virtual 3D environment corresponds to a field of view of the second physical camera in the physical environment; and

projecting the identified object of interest into the virtual 3D environment using the configured first and second virtual cameras.

17. The one or more computer-readable non-transitory storage media of claim 15 , the operations further comprising:

creating a first synthetic depth map for the first virtual camera, wherein the first synthetic depth map provides a one-to-one mapping of each 2D pixel created by the first physical camera and a corresponding space of the first virtual camera in the virtual 3D environment;

creating a second synthetic depth map for the second virtual camera, wherein the second synthetic depth map provides a one-to-one mapping of each 2D pixel created by the second physical camera and a corresponding space of the second virtual camera in the virtual 3D environment; and

creating a camera matrix comprising the first and second synthetic depth maps.

18. The one or more computer-readable non-transitory storage media of claim 17 , wherein projecting the identified object of interest into the virtual 3D environment comprises:

determining a 2D coordinate of the identified object of interest; and

converting the 2D coordinate to a location within the virtual 3D environment using the camera matrix.

19. The one or more computer-readable non-transitory storage media of claim 15 , wherein:

the first virtual camera is placed in the virtual 3D environment at a same latitude, longitude, and altitude as the first physical camera is positioned in the physical environment; and

the second virtual camera is placed in the virtual 3D environment at a same latitude, longitude, and altitude as the second physical camera is positioned in the physical environment.

20. The one or more computer-readable non-transitory storage media of claim 15 , wherein the settings of the first and second virtual cameras and the first and second physical cameras each comprise:

a surge setting;

a sway setting;

a heave setting;

a roll setting;

a pitch setting;

a yaw setting;

a radial distortion; and

a tangential distortion.