Optimizing multi-camera multi-entity artificial intelligence tracking systems
Systems and methods for optimizing multi-camera multi-entity artificial intelligence tracking systems. Visual and location information of entities from video feeds received from multiple cameras can be obtained by employing an entity detection model and re-identification model. Likelihood scores that entity detections belong to an entity track can be predicted from the visual and location information. The entity detections predicted into entity tracks can be processed by employing combinatorial optimization of the likelihood scores by identifying assumptions from the likelihood scores, entity detections, and the entity tracks, filtering the assumptions with unsatisfiable problems to obtain a filtered assumptions set, and optimizing an answer set by utilizing the filtered assumptions set and the likelihood scores to maximize an overall score and obtain optimized entity tracks. Multiple entities can be monitored by utilizing the optimized entity tracks.
1. A computer-implemented method for optimizing multi-camera multi-entity artificial intelligence tracking systems comprising:
obtaining visual information and location information of entities from video feeds received from multiple cameras by employing an entity detection model and a re-identification model;
predicting likelihood scores that entity detections from the visual information and the location information belong to an entity track by employing a multi-layer perceptron;
processing, by a processor device, the entity detections predicted into entity tracks by employing combinatorial optimization of the likelihood scores by:
identifying assumptions from the likelihood scores, the entity detections, and the entity tracks;
filtering the assumptions with unsatisfiable problems to obtain a filtered assumptions set;
optimizing an answer set by utilizing the filtered assumptions set and the likelihood scores to maximize an overall score and obtain optimized entity tracks; and
monitoring multiple entities within a location by utilizing the optimized entity tracks.
2. The computer-implemented method of claim 1 , wherein monitoring multiple entities further comprises tracking multiple patients within a hospital ward.
3. The computer-implemented method of claim 1 , wherein monitoring multiple entities further comprises tracking healthcare providers within a hospital ward.
4. The computer-implemented method of claim 1 , wherein filtering conflicting assumptions further comprises detecting the assumptions with unsatisfiable problems to eliminate.
5. The computer-implemented method of claim 4 , wherein detecting the assumptions further comprises identifying a subset of the assumptions associated with the detected unsatisfiable problems to eliminate.
6. The computer-implemented method of claim 1 , wherein filtering conflicting assumptions further comprises fine-tuning a subset of the assumptions to obtain a conflict-free assumption set.
7. The computer-implemented method of claim 6 , wherein fine-tuning further comprises eliminating an assumption from the subset of the assumptions to obtain a new assumptions set from remaining assumptions from the subset.
8. The computer-implemented method of claim 6 , wherein fine-tuning further comprises obtaining a solution flag of an eliminated assumption by employing answer set programming to a new assumptions set.
9. The computer-implemented method of claim 6 , wherein fine-tuning further comprises processing a reduced subset of the assumptions having another eliminated assumption based on a solution flag.
10. A system for optimizing multi-camera multi-entity artificial intelligence tracking systems comprising:
a memory; and
one or more processor devices in communication with the memory configured to:
obtain visual information and location information of entities from video feeds received from multiple cameras;
predict likelihood scores that entity detections from the visual information and the location information belong to an entity track;
process the entity detections predicted into entity tracks by employing combinatorial optimization of the likelihood scores to:
identify assumptions from the likelihood scores, the entity detections, and the entity tracks;
filter the assumptions with unsatisfiable problems to obtain a filtered assumptions set;
optimize an answer set by utilizing the filtered assumptions set and the likelihood scores to maximize an overall score and obtain optimized entity tracks; and
monitor multiple entities within a location by utilizing the optimized entity tracks.
11. The system of claim 10 , wherein monitoring multiple entities further comprises tracking multiple patients within a hospital ward.
12. The system of claim 11 , wherein monitoring multiple entities further comprises tracking healthcare providers within a hospital ward.
13. The system of claim 10 , wherein filtering conflicting assumptions further comprises detecting the assumptions with unsatisfiable problems to eliminate.
14. The system of claim 13 , wherein detecting the assumptions further comprises identifying a subset of the assumptions associated with the detected unsatisfiable problems to eliminate.
15. The system of claim 10 , wherein filtering conflicting assumptions further comprises fine-tuning a subset of the assumptions to obtain a conflict-free assumption set.
16. The system of claim 15 , wherein fine-tuning further comprises eliminating an assumption from the subset of the assumptions to obtain a new assumptions set from remaining assumptions from the subset.
17. The system of claim 15 , wherein fine-tuning further comprises obtaining a solution flag of an eliminated assumption by employing answer set programming to a new assumptions set.
18. The system of claim 15 , wherein fine-tuning further comprises processing a reduced subset of the assumptions having another eliminated assumption based on a solution flag.
19. A non-transitory computer program product comprising a computer-readable storage medium including program code for optimizing multi-camera multi-entity artificial intelligence tracking systems wherein the program code when executed on a computer causes the computer to perform:
obtaining visual information and location information of entities from video feeds received from multiple cameras by employing an entity detection model and a re-identification model;
predicting likelihood scores that entity detections from the visual information and the location information belong to an entity track by employing a multi-layer perceptron;
processing, by a hardware processor, the entity detections predicted into entity tracks by employing combinatorial optimization of the likelihood scores by:
identifying assumptions from the likelihood scores, the entity detections, and the entity tracks;
filtering the assumptions with unsatisfiable problems to obtain a filtered assumptions set by further:
detecting the assumptions with unsatisfiable problems to eliminate;
identifying a subset of the assumptions associated with the detected unsatisfiable problems;
fine-tuning the subset of the assumptions to obtain a conflict-free assumption set;
optimizing, by an answer set optimization module, an answer set by utilizing the filtered assumptions set and the likelihood scores to maximize an overall score and obtain optimized entity tracks; and
monitoring multiple entities within a location by utilizing the optimized entity tracks.
20. The non-transitory computer program product of claim 19 , wherein monitoring multiple entities further comprises tracking multiple patients within a hospital ward.