Situational awareness-based image annotation systems and methods
Techniques for facilitating situational awareness-based annotation systems and methods are provided. In one example, a method includes receiving a respective image from each of a plurality of imaging systems. The method further includes generating a panoramic image based on the images received from the plurality of imaging devices. The method further includes detecting an object in the panoramic image. The method further includes determining, for each image received from the plurality of imaging systems, whether the image contains the object detected in the panoramic image. The method further includes annotating each image received from the plurality of imaging systems determined to contain the object to include an indication associated with the object in the image. Related systems are also provided.
1. A method comprising:
receiving a respective image from each of a plurality of imaging systems;
generating a single-first panoramic image based on the images received from the plurality of imaging systems;
detecting an object in the single panoramic image, wherein the detecting comprises performing an object detection process on the single panoramic image without, prior to the performing the object detection process on the single panoramic image, performing a separate object detection process on each image received from the plurality of imaging systems and used to generate the single panoramic image;
determining, for each image received from the plurality of imaging systems and used to generate the single panoramic image, whether the image contains the object detected in the single panoramic image; and
annotating each image received from the plurality of imaging systems and determined to contain the object to include an indication associated with the object in the image.
2. The method of claim 1 , further comprising determining, for each image received from the plurality of imaging systems and determined to contain the object, a location of the object in the image based at least on a location of the object in the single panoramic image and overlap associated with the images received from the plurality of imaging systems, wherein the detecting is based at least on convolutional neural networks, and wherein the overlap associated with the images is based on an overlap between field of views of the plurality of imaging systems.
3. The method of claim 1 , further comprising:
receiving a respective second image from each of a second plurality of imaging systems;
generating a second panoramic image based on the second images received from the second plurality of imaging systems;
detecting a second object in the second panoramic image;
determining, for each second image received from the second plurality of imaging systems and used to generate the second panoramic image, whether the second image contains the second object detected in the second panoramic image; and
annotating each second image received from the second plurality of imaging systems and determined to contain the second object to include an indication associated with the second object in the second image,
wherein:
at least one of the images from the plurality of imaging systems is a panoramic image having a field of view narrower than the single panoramic image;
the single panoramic image is an infrared panoramic image; and
the second panoramic image is a visible-light panoramic image.
4. The method of claim 3 , further comprising:
receiving a respective stream of video data from each of the plurality of imaging systems, wherein the respective image is received as a part of the respective stream of video data; and
receiving a respective stream of video data from each of the second plurality of imaging systems, wherein the respective second image is received as a part of the respective stream of video data, and wherein each stream of video data is a real-time streaming protocol (RTSP) stream.
5. The method of claim 1 , wherein the single panoramic image provides a 360 degree field of view, and wherein the method further comprises for each imaging system of the plurality of imaging systems:
capturing, by the imaging system, a respective plurality of images, wherein each image of the respective plurality of images is associated with a respective portion of the 360 degree field of view; and
generating, by the imaging system, the image based on the respective plurality of images.
6. The method of claim 1 , wherein the determining and the annotating are performed after the generating and the detecting, and wherein the image from one of the plurality of imaging systems has a portion that overlaps with the image from another of the plurality of imaging systems.
7. The method of claim 1 , further comprising:
receiving a respective second image from each of a second plurality of imaging systems; and
determining, for each second image received from the second plurality of imaging systems, whether the second image contains the object detected in the single panoramic image based on the detecting and the determining, for each image received from the plurality of imaging systems, whether the image contains the object.
8. The method of claim 1 , further comprising:
receiving, by a display device, each annotated image as part of a respective stream of video data; and
displaying, by the display device, each annotated image.
9. A system comprising:
a plurality of imaging systems, wherein each of the plurality of imaging systems is configured to capture a respective image; and
a processing circuit comprising:
a combiner circuit configured to:
receive the images from the plurality of imaging systems; and
generate a single panoramic image based on the images received from the plurality of imaging systems;
a detection circuit configured to detect an object in the single panoramic image by performing an object detection process on the single panoramic image without, prior to the performing the object detection process on the single panoramic image, performing a separate object detection process on each image received from the plurality of imaging systems and used to generate the single panoramic image;
an annotation circuit configured to:
determine, for each image received from the plurality of imaging systems and used by the combiner circuit to generate the single panoramic image, whether the image contains the object detected in the single panoramic image;
determine, for each image received from the plurality of imaging systems and determined to contain the object, a location of the object in the image based at least on a location of the object in the single panoramic image; and
annotate each image received from the plurality of imaging systems and determined by the annotation circuit to contain the object to include an indication associated with the object in the image, wherein a location of the indication in the annotated image is based on a location of the object in the image.
10. The system of claim 9 , wherein the annotation circuit is configured to determine, for each image received from the plurality of imaging systems and determined to contain the object, the location of the object in the image further based on overlap associated with the images received from the plurality of imaging systems, wherein the combiner circuit is configured to detect the object based on convolutional neural networks, and wherein, for each image received from the plurality of imaging systems and determined to contain the object, the indication associated with the object comprises a bounding box positioned around the object in the image.
11. The system of claim 9 , further comprising a second plurality of imaging systems, wherein each of the second plurality of imaging systems is configured to capture a respective second image;
wherein the combiner circuit is further configured to:
receive the second images from the second plurality of imaging systems; and
generate a second panoramic image based on the second images received from the second plurality of imaging systems;
wherein the detection circuit is further configured to detect a second object in the second panoramic image;
wherein the annotation circuit is further configured to:
determine, for each second image received from the second plurality of imaging systems and used by the combiner circuit to generate the second panoramic image, whether the second image contains the second object detected in the second panoramic image; and
annotate each second image received from the second plurality of imaging systems and determined by the annotation circuit to contain the second object to include an indication associated with the second object in the second image.
12. The system of claim 11 , wherein:
the plurality of imaging systems comprises a plurality of four thermal infrared imaging systems;
the second plurality of imaging systems comprises a plurality of four visible-light imaging systems;
the single panoramic image is an infrared panoramic image;
the second panoramic image is a visible-light panoramic image; and
the system further comprises a plurality of multiplexer circuits, wherein each of the plurality of multiplexer circuits is configured to selectively provide, to the combiner circuit, the image from a respective one of the four thermal infrared imaging systems or a respective one of the four visible-light imaging systems.
13. The system of claim 12 , wherein:
each of the infrared panoramic image and the visible-light panoramic image provides a 360 degree field of view;
each of the four thermal infrared imaging systems comprises:
a plurality of thermal infrared imaging devices, wherein each of the plurality of thermal infrared imaging devices is configured to capture a respective thermal image associated with a portion of the 360 degree field of view; and
a stitching circuit configured to generate the image of the thermal infrared imaging system based on the thermal images captured by the plurality of thermal infrared imaging devices; and
each of the four visible-light imaging systems comprises:
a plurality of visible-light imaging devices, wherein each of the plurality of visible-light imaging devices is configured to capture a respective visible-light image associated with a portion of the 360 degree field of view; and
a stitching circuit configured to generate the image of the visible-light imaging system based on the visible-light images captured by the plurality of visible-light imaging devices.
14. The system of claim 9 , wherein at least one of the images from the plurality of imaging systems is a panoramic image having a field of view narrower than the single panoramic image, wherein the combiner circuit is further configured to receive the respective image from each of the plurality of imaging systems as a part of a respective stream of video data from each of the plurality of imaging systems, and wherein each stream of video data is a real-time streaming protocol (RTSP) stream.
15. The system of claim 9 , wherein the plurality of imaging systems are collectively configured to capture a 360 degree field of view, wherein each of the plurality of imaging systems is configured to capture a portion of the 360 degree field of view, and wherein the image from one of the plurality of imaging systems has a portion that overlaps with the image from another of the plurality of imaging systems.
16. The system of claim 9 , further comprising a display device configured to:
receive each annotated image as part of a respective stream of video data; and
display each annotated image.
17. The system of claim 9 , further comprising a plurality of radar systems configured to provide a 360 degree composite field of view and capture radar data in the 360 degree composite field of view.
18. A vehicle comprising the system of claim 9 , wherein each of the plurality of imaging systems is attached to the vehicle.
19. The vehicle of claim 18 , wherein the plurality of imaging systems are configured to collectively provide a 360 degree field of view around the vehicle.
20. The system of claim 9 , wherein, for each image received from the plurality of imaging systems and determined to contain the object, the location of the object in the image is further based on overlap associated with the images received from the plurality of imaging systems.