Guided multi-spectral inspection
An imaging system is provided. A first imaging system captures initial sensor data in a form of visible domain data. A second imaging system captures subsequent sensor data in a form of second domain data, wherein the initial and subsequent sensor data are of different spectral domains. A controller subsystem detects at least one region of interest in real-time by applying a machine learning technique to the visible domain data, localizes at least one object of interest in the at least one region of interest to generate positional data for the at least one object of interest, and autonomously steers a point of focus of the second imaging system to a region of a scene including the object of interest to capture the second domain data responsive to the positional data.
1. An imaging system, comprising:
a first imaging system that captures initial sensor data in a form of visible domain data;
a second imaging system, on a vehicle, that captures subsequent sensor data in a form of second domain data, wherein the initial and the subsequent sensor data are of different spectral domains; and
a controller subsystem, operatively coupled to the first and second imaging systems, that
detects at least one region of interest in a scene in real-time by applying a soft visual attention based neural network trained using attention weights to focus on regions of interest and pre-defined types of objects of interest in the visible domain data from the first imaging system,
localizes, using the soft visual attention based neural network, at least one object of interest in the at least one region of interest to generate positional data for the at least one object of interest, and
autonomously deploys the vehicle to the scene and steers a point of focus of the second imaging system to a region of the scene including the object of interest to capture the second domain data responsive to the positional data,
wherein the first and the second imaging systems share a common set of x and y coordinates for any captured scenes, the positional data being generated based on the common set, and being used to control an effective field of view (FoV) of the second imaging system, the controller subsystem being further configured for dynamically adjusting an effective Field of View (FoV) of the second imaging system based on real-time coordinates changes detected through the first imaging system.
2. The imaging system of claim 1 , further comprising autonomously providing depth information and material properties of the at least one object.
3. The imaging system of claim 1 , wherein the first imaging system comprises an imaging device selected from the group consisting of an RGB camera, an InfraRed (IR) camera, and a thermal camera.
4. The imaging system of claim 3 , wherein the second imaging system comprises a radar imaging system.
5. The imaging system of claim 3 , wherein the second imaging system comprises a mm Wave imaging system.
6. The imaging system of claim 1 , wherein localize the at least one object of interest comprises extracting coordinates of the at least one object of interest.
7. The imaging system of claim 1 , wherein the second domain data includes information from the scene unobservable in the visible domain.
8. The imaging system of claim 1 , further comprising controlling an autonomously driven motor vehicle responsive to the subsequent sensor data.
9. The imaging system of claim 1 , wherein the first imaging system is on an unmanned aerial vehicle (UAV), and the first imaging system comprises an imaging device selected from the group consisting of an RGB camera, an InfraRed (IR) camera, and a thermal camera.
10. The imaging system of claim 1 , wherein the first imaging system is a thermal camera for detecting temperature hot spots in the scene as the at least one region of interest.
11. The imaging system of claim 1 , wherein the controller subsystem is configured in a cloud computing configuration.
12. A method for imaging, comprising:
capturing, by a first imaging system, initial sensor data in a form of visible domain data;
capturing, by a second imaging system on a vehicle, subsequent sensor data in a form of second domain data, the initial and the subsequent sensor data being of different spectral domains;
detecting, by a controller subsystem, at least one region of interest in a scene in real-time by applying a soft visual attention based neural network trained using attention weights to focus on regions of interest and pre-defined types of objects of interest in the visible domain data from the first imaging system;
localizing, by the controller subsystem using the soft visual attention based neural network, at least one object of interest in the at least one region of interest to generate positional data for the at least one object of interest; and
autonomously deploying, by the controller subsystem, the vehicle to the scene and steering a point of focus of a second imaging system to the scene to capture subsequent sensor data in a form of second domain data responsive to the positional data,
wherein the initial and the subsequent sensor data are of different spectral domains, the first and the second imaging systems sharing a common set of x and y coordinates for any captured scenes, the positional data being generated based on the common set, and being used to control an effective field of view (FoV) of the second imaging system, the controller subsystem being further configured for dynamically adjusting an effective Field of View (FoV) of the second imaging system based on real-time coordinates changes detected through the first imaging system.
13. The method of claim 12 , wherein the second domain data includes information from the scene unobservable in the visible domain.
14. The method of claim 12 , further comprising autonomously providing depth information and material properties of the at least one object.
15. The method of claim 12 , wherein localize the at least one object of interest comprises extracting coordinates of the at least one object of interest.
16. The method of claim 12 , further comprising controlling an autonomously driven motor vehicle responsive to the subsequent sensor data.
17. A computer program product for imaging, the computer program product comprising a non-transitory computer readable storage medium having program instructions embodied therewith, the program instructions executable by a computing system to cause the computing system to perform a method, the method comprising:
capturing, by a first imaging system of the computing system, initial sensor data in a form of visible domain data;
detecting, by a controller subsystem of the computing system, at least one region of interest of a scene in real-time by applying a soft visual attention based neural network trained using attention weights to focus on regions of interest and pre-defined types of objects of interest in the visible domain data from the first imaging system;
localizing, by the controller subsystem using the soft visual attention based neural network, at least one object of interest in the at least one region of interest to generate positional data for the at least one object of interest; and
autonomously deploying, by the controller subsystem, a vehicle having a second imaging system and steering a point of focus of the second imaging system to the scene to capture subsequent sensor data in a form of second domain data responsive to the positional data,
wherein the initial and the subsequent sensor data are of different spectral domains, the first and the second imaging systems sharing a common set of x and y coordinates for any captured scenes, the positional data being generated based on the common set, and being used to control an effective field of view (FoV) of the second imaging system, the controller subsystem being further configured for dynamically adjusting an effective Field of View (FoV) of the second imaging system based on real-time coordinates changes detected through the first imaging system.