System and method for selecting a part of a video image for a face detection operation
A cropped bounding box selection operation is performed on a video captured by a video capture and playback system, to select one or more cropped bounding boxes from the video for processing by a face detection operation. The cropped bounding box selection operation identifies objects from the video images and assigns a ranking to each identified object based on certain priority criteria; one or more cropped bounding boxes corresponding to the objects with the highest ranking(s) are then processed by the face detection operation to detect a face in each object.
1. A non-transitory computer readable medium having recorded thereon program code executable by a processor to perform a method comprising:
receiving a video frame of a scene captured by a video capture device;
identifying multiple objects in the video frame and producing multiple cropped bounding boxes each comprising a respective portion of the video frame bounding at least a respective portion of one of the identified objects and metadata representing one or more characteristics of the identified objects or cropped bounding boxes;
assigning a respective ranking of a plurality of rankings, based on associating the one or more characteristics with a higher or lower priority for detecting a human face in a respective cropped bounding box of the cropped bounding boxes, to each of the identified objects, wherein the respective ranking is a respective product of a respective characteristic multiplier, being a function of relevance of a particular characteristic to face detection, and a respective characteristic weighting factor;
producing a priority table comprising the identified objects listed by the assigned rankings; and
selecting a subset of the multiple cropped bounding boxes for a face detection operation,
wherein inclusion in the selected subset is based on the respective one or more rankings of one or more of the identified objects from the priority table selected for inclusion in the selected subset.
2. The non-transitory computer readable medium as claimed in claim 1 , wherein the one or more characteristics include a trajectory, and wherein a trajectory moving away from the video capture device is associated with the lower priority.
3. The non-transitory computer readable medium as claimed in claim 2 , further comprising identifying an occlusion in the video frame, and wherein a trajectory moving towards the occlusion is associated with the higher priority.
4. The non-transitory computer readable medium as claimed in claim 1 , wherein the one or more characteristics include a human classification, and wherein a human classification having a confidence value exceeding a selected threshold is associated with the higher priority.
5. The non-transitory computer readable medium as claimed in claim 1 , wherein the one or more characteristics include a color, and wherein a color corresponding to a human skin tone is associated with the higher priority.
6. The non-transitory computer readable medium as claimed in claim 5 , wherein the one or more characteristics include pixel location and color, and wherein a selected number of contiguous pixels having colors corresponding to a human skin tone is associated with the higher priority.
7. The non-transitory computer readable medium as claimed in claim 1 , wherein the one or more characteristics include pixel sharpness or contrast.
8. The non-transitory computer readable medium as claimed in claim 1 , wherein the one or more characteristics is a previously detected face, and wherein an object having a previously detected face detected after a selected time period is associated with the higher priority.
9. The non-transitory computer readable medium as claimed in claim 1 , wherein the one or more characteristics further include a quality of a previously detected face, and wherein an object having a previously detected face with a quality below a selected threshold is associated with the higher priority.
10. The non-transitory computer readable medium as claimed in claim 1 , wherein the one or more characteristics includes a location, and wherein a location with a number of previous face detections exceeding a selected threshold is associated with the higher priority.
11. A method for selecting an object for a face detection operation, comprising:
receiving a video frame of a scene captured by a video capture device;
identifying multiple objects in the video frame and producing multiple cropped bounding boxes each comprising a respective portion of the video bounding at least a respective portion of one of the identified objects and metadata representing one or more characteristics of the identified objects or cropped bounding boxes;
assigning a respective ranking of a plurality of rankings, based on the one or more characteristics being associated with a higher or lower priority for detecting a human face in a respective cropped bounding box of the cropped bounding boxes, to each of the identified objects, wherein the respective ranking is a respective product of a respective characteristic multiplier, being a function of relevance of a particular characteristic to face detection, and a respective characteristic weighting factor;
producing a priority table comprising the identified objects listed by the assigned rankings; and
selecting a subset of the multiple cropped bounding boxes for the face detection operation, and
wherein inclusion in the selected subset is based on the respective one or more rankings of one or more of the identified objects from the priority table selected for inclusion in the selected subset.
12. The method as claimed in claim 11 , wherein the one or more characteristics include a trajectory, and wherein a trajectory moving away from the video capture device is associated with the lower priority.
13. The method as claimed in claim 12 , further comprising identifying an occlusion in the video frame, and wherein a trajectory moving towards the occlusion is associated with the higher priority.
14. The method as claimed in claim 11 , wherein the one or more characteristics include a human classification, and wherein a human classification having a confidence value exceeding a selected threshold is associated with the higher priority.
15. The method as claimed in claim 11 , wherein the one or more characteristics include a color, and wherein a color corresponding to a human skin tone is associated with the higher priority.
16. The method as claimed in claim 15 , wherein the one or more characteristics include pixel location and color, and wherein a selected percentage of pixels having colors corresponding to a human skin tone is associated with the higher priority.
17. The method as claimed in claim 11 , wherein the one or more characteristics include pixel sharpness or contrast.
18. The method as claimed in claim 11 , wherein the one or more characteristics is a previously detected face, and wherein an object having a previously detected face detected after a selected time period is associated with the higher priority.
19. The method as claimed in claim 11 , wherein the one or more characteristics further include a quality of a previously detected face, and wherein an object having a previously detected face with a quality below a selected threshold is associated with the higher priority.
20. The method as claimed in claim 11 , wherein the one or more characteristics includes a location, and wherein a location with a number of previous face detections exceeding a selected threshold is associated with the higher priority.
21. A system for selecting an object for a face detection operation, comprising:
at least one video capture device configured to capture at least one video of a scene; and
a video analytics module communicative with the at least one video capture device to receive the at least one video, and comprising a processor and a memory having stored thereon program code executable by the processor to:
identify multiple objects in a video frame of the at least one video and produce cropped bounding boxes each comprising a respective portion of the least one video frame bounding at least a respective portion of one of the identified objects and metadata representing one or more characteristics of the identified objects or cropped bounding boxes;
assign a respective ranking of a plurality of rankings, based on the one or more characteristics being associated with a higher or lower priority for detecting a human face in a respective cropped bounding box of the cropped bounding boxes, to each of the identified objects, wherein the respective ranking is a respective product of a characteristic multiplier, being a function of relevance of a particular characteristic to face detection, and a respective characteristic weighting factor;
producing a priority table comprising the identified objects listed by the assigned rankings; and
select a subset of the multiple cropped bounding boxes for the face detection operation, and
wherein inclusion in the selected subset is based on the respective one or more rankings of one or more of the identified objects from the priority table selected for inclusion in the selected subset.