System and method for tracking and recognizing people

Abstract

A tracking and recognition system is provided. The system includes a computer vision-based identity recognition system configured to recognize one or more persons, without a priori knowledge of the respective persons, via an online discriminative learning of appearance signature models of the respective persons. The computer vision-based identity recognition system includes a memory physically encoding one or more routines, which when executed, cause the performance of constructing pairwise constraints between the unlabeled tracking samples. The computer vision-based identity recognition system also includes a processor configured to receive unlabeled tracking samples collected from one or more person trackers and to execute the routines stored in the memory via one or more algorithms to construct the pairwise constraints between the unlabeled tracking samples.

1 . A tracking and recognition system, comprising:
 a computer vision-based identity recognition system configured to recognize one or more persons, without a priori knowledge of the respective persons, via an online discriminative learning of appearance signature models of the respective persons, wherein the computer vision-based identity recognition system comprises:
 a memory physically encoding one or more routines, which when executed, cause the performance of constructing pairwise constraints between unlabeled tracking samples; and 
 a processor configured to receive unlabeled tracking samples collected from one or more person trackers and to execute the routines stored in the memory via one or more algorithms to construct the pairwise constraints between the unlabeled tracking samples. 
   
     
     
         2 . The system of  claim 1 , wherein the pairwise constraints comprise a must-link constraint between two tracking samples from a single tracker. 
     
     
         3 . The system of  claim 1 , wherein the pairwise constraints comprise a cannot-link constraint between two tracking samples from different trackers. 
     
     
         4 . The system of  claim 1 , wherein the routines, when executed, cause the performance of weighing each pairwise constraint between the unlabeled tracking samples. 
     
     
         5 . The system of  claim 4 , wherein the routines, when executed, cause the performance of spectral clustering of the unlabeled tracking samples with weighted pairwise constraints. 
     
     
         6 . The system of  claim 5 , wherein the routines, when executed, utilize a kernel learning based function to spectral cluster the unlabeled tracking samples with weighted pairwise constraints. 
     
     
         7 . The system of  claim 5 , wherein the routines, when executed, cause the performance of learning a respective appearance signature model for each cluster of unlabeled tracking samples. 
     
     
         8 . The system of  claim 7 , wherein the respective appearance signature model comprises a new appearance signature model. 
     
     
         9 . The system of  claim 7 , wherein the respective appearance signature model comprises an updated appearance signature model. 
     
     
         10 . The system of  claim 7 , wherein the routines, when executed, utilize a support vector machine to learn the respective appearance signature model for each cluster of unlabeled tracking samples. 
     
     
         11 . The system of  claim 1 , wherein the processor is configured to receive and buffer the unlabeled tracking samples in an online and asynchronous mode. 
     
     
         12 . The system of  claim 1 , wherein the one or more person trackers comprise 3D ground plane-based trackers maintained in real-time. 
     
     
         13 . The system of  claim 1 , wherein the unlabeled tracking samples comprise noisy samples having spatiotemporal gaps. 
     
     
         14 . A method for tracking and recognition of people, comprising:
 generating tracking samples from one or more person trackers of a tracking system;   receiving unlabeled tracking samples from the generated tracking samples into a data buffer for a time span;   generating weighted pairwise constraints between the unlabeled tracking samples;   generating clusters via spectral clustering of the unlabeled tracking samples with weighted pairwise constraints; and   learning in an online and discriminative manner a respective appearance signature model for each respective cluster.   
     
     
         15 . The method of  claim 14 , wherein learning the respective appearance signature model for each respective cluster comprises learning in an online and discriminative manner. 
     
     
         16 . The method of  claim 14 , wherein the one or more person trackers comprise 3D ground plane-based trackers maintained in real-time, and generating tracking samples comprises extracting projected image regions from the 3D ground plane-based trackers. 
     
     
         17 . The method of  claim 14 , receiving the unlabeled tracking samples, via batch processing, in an online and asynchronous mode. 
     
     
         18 . The method of  claim 14 , wherein the data buffer reaching a threshold size from the received unlabeled tracking samples activates the generation of the weighted pairwise constraints between the unlabeled tracking samples and the clusters and the online discriminative learning of the respective appearance signature model for each respective cluster. 
     
     
         19 . The method of  claim 14 , wherein a portion of the received unlabeled tracking samples in the data buffer overlap from two successive time spans. 
     
     
         20 . The method of  claim 14 , wherein the weighted pairwise constraints comprise a must-link constraint between two tracking samples from a single tracker and a cannot-link constraint between two tracking samples from different trackers. 
     
     
         21 . The method of  claim 14 , wherein the respective appearance signature model comprises a new appearance signature model or an updated appearance signature model. 
     
     
         22 . A non-transitory, computer-readable media comprising one or more routines which executed by at least one processor causes acts to be performed comprising:
 receiving unlabeled tracking samples collected from one or more person trackers;   generating weighted pairwise constraints between the unlabeled tracking samples;   generating clusters via spectral clustering of the unlabeled tracking samples with weighted pairwise constraints; and   learning in an online and discriminative manner a respective appearance signature model for each respective cluster.   
     
     
         23 . The non-transitory, computer-readable media of  claim 22 , wherein the weighted pairwise constraints comprise a must-link constraint between two tracking samples from a single tracker and a cannot-link constraint between two tracking samples from different trackers. 
     
     
         24 . The non-transitory, computer readable media of  claim 23 , wherein processor utilizes a multi-class support vector machine to learn the respective appearance signature model for each respective cluster. 
     
     
         25 . The non-transitory, computer readable media of  claim 24 , wherein the multi-class support vector machine comprises an incremental support vector machine that continuously updates itself upon receiving new data.