SYSTEM AND METHOD FOR PROCESSING DIGITAL MEDIA
First Claim
1. A method to identify an unknown object extracted from a video stream, the method comprising:
- detecting the unknown object in a video frame, the video frame being one of a plurality of video frames from the video stream;
receiving a plurality of inputs, each input in the plurality of inputs being a probable identity of the unknown object in the video frame;
associating each input with the unknown object;
associating a relative weight with each input, the relative weight being a likelihood of an input to match the unknown object;
obtaining values of a distance from each probable identity associated with the unknown object to each reference object;
obtaining a joint distance between every unknown object and each reference object, the joint distance being a weighted transformation of distances from each probable identity associated with the unknown object to reference object;
comparing the joint distances or input modality distances from the unknown object to each reference object; and
identifying the unknown object based on the comparing.
5 Assignments
0 Petitions
Accused Products
Abstract
A method for processing digital media is described. In one example embodiment, the method may include detecting an unknown object in a video frame, receiving inputs representing probable identities of the unknown object in the video frame from various sources, and associating each input with the unknown object detected in the video frame. The received inputs may be processed, compared with reference data and, based on the comparison, probable identities of the object associated with the input derived. The method may further include retrieving a likelihood of the input to match the unknown object from historical data and producing weights corresponding to the inputs, fusing the inputs and the relative weight associated with each input, and identifying the unknown object based on a comparison of the weighted distances from the unknown identify to a reference identity. The relative weights are chosen from the historical data to maximize correct recognition rate based on the history of recognitions and manual verification results.
67 Citations
25 Claims
-
1. A method to identify an unknown object extracted from a video stream, the method comprising:
-
detecting the unknown object in a video frame, the video frame being one of a plurality of video frames from the video stream; receiving a plurality of inputs, each input in the plurality of inputs being a probable identity of the unknown object in the video frame; associating each input with the unknown object; associating a relative weight with each input, the relative weight being a likelihood of an input to match the unknown object; obtaining values of a distance from each probable identity associated with the unknown object to each reference object; obtaining a joint distance between every unknown object and each reference object, the joint distance being a weighted transformation of distances from each probable identity associated with the unknown object to reference object; comparing the joint distances or input modality distances from the unknown object to each reference object; and identifying the unknown object based on the comparing. - View Dependent Claims (2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15)
-
-
16. A system to identify an unknown object extracted from a video stream, the system comprising:
-
a buffered frame sequence processor to process a plurality of video frames comprising the video stream; a facial context extraction processor to detect and extract the unknown object in a video frame, the video frame being one of the plurality of video frames comprising the video stream; a receiving module to receive a plurality of inputs, each input being a predicted identity of the unknown object in the video frame; an associating module to associate each input with the unknown object detected in the video frame and to associate a relative weight with each input, the relative weight being a likelihood of an input to correctly identify the unknown object; a computing module to obtain values of a distance function from the unknown object in the video frame and each reference object selected from a set of known objects based on the plurality of the inputs and the relative weight associated with each input; a comparing module to compare the values of the distance function for the unknown object and each reference object; and an identification module to identify the unknown object based on the comparing. - View Dependent Claims (17, 18, 19, 20, 21, 22, 23)
-
-
24. A method to identify an unknown object extracted from a video stream, the method comprising;
-
means of detecting the unknown object in a video frame, the video frame being one of a plurality of video frames from the video stream; means of receiving a plurality of inputs, each input in the plurality of the inputs being a probable identity of the unknown object in the video frame; means of associating each input with the unknown object; means associating a relative weight with each input, the relative weight being a likelihood of an input to match the unknown object; means of obtaining values of a distance from each probable identity associated with the unknown object to each reference object; means of obtaining a joint distance between every unknown object and each reference object, the joint distance being a weighted transformation of distances from each probable identity associated with the unknown object to reference object; means of comparing the joint distances or input modality distances from the unknown object to each reference object; and means of identifying the unknown object based on the comparing.
-
-
25. A machine-readable medium comprising instructions, which when implemented by one or more processors perform the following operations:
-
detect the unknown object in a video frame, the video frame being one of a plurality of video frames from the video stream; receive a plurality of inputs, each input in the plurality of the inputs being a probable identity of the unknown object in the video frame; associate each input with the unknown object; associate a relative weight with each input, the relative weight being a likelihood of an input to match the unknown object; obtain values of a distance from each probable identity associated with the unknown object to each reference object; obtain a joint distance between every unknown object and each reference object, the joint distance being a weighted transformation of distances from each probable identity associated with the unknown object to reference object; compare the joint distances or input modality distances from the unknown object to each reference object; and identify the unknown object based on the comparing.
-
Specification