Primary user selection for head tracking
地域：
NV
NV Reno
摘要
Various embodiments enable a primary user to be identified and tracked using stereo association and multiple tracking algorithms. For example, a face detection algorithm can be run on each image captured by a respective camera independently. Stereo association can be performed to match faces between cameras. If the faces are matched and a primary user is determined, a face pair is created and used as the first data point in memory for initializing object tracking. Further, features of a user's face can be extracted and the change in position of these features between images can determine what tracking method will be used for that particular frame.
說明書
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
People are increasingly interacting with computers and other electronic devices in new and interesting ways. For example, object tracking can be implemented for recognizing certain user gestures, such as head nods or shakes, eye winks or other ocular motion, or hand and/or finger gestures, as input for the device. Object tracking can also be utilized for advanced device security features such as ensuring “l(fā)ive” facial recognition, fingerprinting, retinal scanning, or identification based on gait. Devices capable of object tracking can also be configured for video editing techniques such as video stabilization (e.g., to remove jitter) or to render smooth camera motions due to panning, tilting, or dollying in/dollying out. There are, however, many challenges to properly tracking an object due to, for example, abrupt motions, changes in appearance or background, device motion, among others. Further, factors such as image sensor and lens characteristics, illumination conditions, noise, and occlusion can also affect how an object is represented from image to image or frame to frame. Additionally, the processing requirements for adequate object tracking can often be at odds with the objective of minimizing processing and power use on portable computing devices.
Various embodiments in accordance with the present disclosure will be described with reference to the drawings, in which:
權(quán)利要求
What is claimed is:1. A computing device, comprising:at least one processor; a right camera having a first field of view; a left camera having a second field of view at least partially overlapping the first field of view; memory including instructions that, when executed by the at least one processor, cause the computing device to: capture right image data with the right camera; capture left image data with the left camera; detect, using a face detection algorithm, a first right representation of a first face and a second right representation of a second face in the right image data, the face detection algorithm returning a first right bounding box for the first right representation and a second right bounding box for the second right representation; detect, using the face detection algorithm, a first left representation of the first face and a second left representation of the second face in the left image data, the face detection algorithm returning a first left bounding box for the first face and a second left bounding box for the second face, wherein the first right representation and the first left representation form a first set of representations for the first face and the second left representation and the second right representation form a second set of representations for the second face; determine that the first set of representations satisfy a first selection criterion, the first selection criterion corresponds to a larger stereo disparity; determine that the first set of representations satisfy a second selection criterion, the second selection criterion corresponds to a larger average bounding box size; determine that the first set of representations satisfy a third selection criterion, the third selection criterion corresponds to a shorter average distance to center of a camera field of view; and select the first face as a primary face for a face tracking algorithm to track in subsequent images based on at least one of the first set of representations satisfying two or more of the first selection criterion, the second selection criterion, or the third selection criterion. 2. The computing device of claim 1 , wherein the instructions when executed further cause the computing device to:determine, based at least in part on the first stereo disparity, a first length between a first right eye and a first left eye of the first face; determine the first length is greater or less than an acceptable range of eye lengths; and determine the first face is not a user based on the first length being greater or less than the acceptable range of eye lengths. 3. The computing device of claim 1 , wherein the instructions when executed further cause the computing device to:determine a first stereo disparity associated with the first face by calculating a difference between a first position of a first center of the first right bounding box and a second position of a second center of the first left bounding box. 4. A computer-implemented method, comprising:acquiring image data captured by one or more cameras; detecting one or more first bounding boxes associated with a first object represented in the image data; detecting one or more second bounding boxes associated with a second object represented in the image data; determining a first size associated with the one or more first bounding boxes; assigning a first value associated with the first size to the first object; determining a second size associated with the one or more second bounding boxes; determining a first distance associated with the one or more cameras and the one or more first bounding boxes; assigning a second value associated with the first distance to the first object; determining a second distance associated with the one or more cameras and the one or more second bounding boxes; determining a third distance associated with a reference point in the image data and the one or more first bounding boxes; assigning a third value associated with the third distance to the first object; determining a first combination value associated with the first object based at least in part by applying the first value the second value and the third value; determining a second combination value associated with the second object; and selecting the first object as a primary object based at least in part on the first combination value and the second combination value. 5. The computer-implemented method of claim 4 , further comprising:detecting a first bounding box associated with the first object represented in a first image of the image data; detecting a second bounding box associated with the first object represented in a second image of the image data; and determining the first distance as a displacement between the first bounding box and the second bounding box. 6. The computer-implemented method of claim 4 , further comprising:determining a first stereo disparity between a first representation of the first object in a first image of the image data and a second representation of the first object in a second image of the image data; and determining a second stereo disparity between a third representation of the second object in the first image and a fourth representation of the second object in the second image, wherein the first object is a first face and the second object is a second face. 7. The computer-implemented method of claim 6 , further comprising:determining, based at least in part on the first stereo disparity, a first length between a first right eye and a first left eye of the first face; determining, based at least in part on the second stereo disparity, a second length between a second right eye and a second left eye of the second face; determining at least one of the first length or the second length is greater or less than an acceptable range of eye lengths; and determining at least one of the first face or the second face is not a user based on at least one of the first length or the second length being greater or less than the acceptable range of eye lengths. 8. The computer-implemented method of claim 6 , further comprising:determining, based at least in part upon the first stereo disparity, a fifth distance between the first face and a display screen; and determining, based at least in part upon the second stereo disparity, a sixth distance between the second face and the display screen, wherein selecting the primary object is further based at least in part on a shortest distance between the fifth distance and the sixth distance. 9. The computer-implemented method of claim 8 , further comprising:detecting the one or more first bounding boxes from first pixels associated with the first face in the image data; and determining the one or more second bounding boxes from second pixels associated with the second face in the image data. 10. The computer-implemented method of claim 9 , further comprising:determining first features for the first face using a feature extraction algorithm; and determining second features for the second face using the feature extraction algorithm. 11. The computer-implemented method of claim 4 , further comprising:detecting a first face represented in a first image of the image data as a first representation of the first object; detecting the first face represented in a second image of the image data as a second representation of the first object; determining that the first representation matches the second representation; associating the first representation with the second representation; detecting a second face represented in the first image as a third representation of the second object; detecting the second face represented in the second image as a fourth representation of the second object; determining that the third representation matches the fourth representation; and associating the third representation with the fourth representation. 12. The computer-implemented method of claim 5 , further comprising:capturing the first image using a first camera of the one or more cameras having a first field of view; and capturing a second image of the image data using a second camera of the one or more cameras having a second field of view that at least partially overlapping the first field of view. 13. A computing device, comprising:at least one processor; memory including instructions that, when executed by the at least one processor, cause the computing device to:acquire image data captured by one or more cameras; detect one or more first bounding boxes associated with a first face represented in the image data; detect one or more second bounding boxes associated with a second face in the image data; determine a first size associated with the one or more first bounding boxes; assign a first value associated with the first size to the first face; determine a second size associated with the one or more second bounding boxes; determine a first distance associated with the one or more cameras and the one or more first bounding boxes; assign a second value associated with the first distance to the first face; determine a second distance associated with the one or more cameras and the one or more second bounding boxes; determine a third distance associated with a reference point in the image data and the one or more first bounding boxes; assign a third value associated with the third distance to the first face; determine a first combination value associated with the first face based at least in part by applying the first value, the second value, and the third value; determine a second combination value associated with the second face; and select the first face as a primary face based at least in part on determining the first combination value exceeds the second combination value. 14. The computing device of claim 13 , wherein the instructions when executed further cause the computing device to:detect a first bounding box associated with the first face represented in a first image of the image data; detect a second bounding box associated with the first face represented in a second image of the image data; and determine the first distance as a displacement between the first bounding box and the second bounding box. 15. The computing device of claim 13 , wherein the instructions when executed further cause the computing device to:determine a fourth distance associated with the reference point and the one or more second bounding boxes assign a fourth value associated with the second size to the second face; assign a fifth value associated with the second distance to the second face; assign a sixth value associated with the fourth distance to the second face; and determine the second combination value associated with the second face based at least in part by applying the fourth value, the fifth value, and the sixth value. 16. The computing device of claim 13 , wherein the instructions when executed further cause the computing device to:determine a first stereo disparity between a first bounding box associated with the first face represented in a first image of the image data and a second bounding box associated with the first face in a second image of the image data; and determine a second stereo disparity between a third bounding box associated with the second face represented in the first image and a fourth bounding box associated with the second face represented in the second image. 17. The computing device of claim 16 , wherein the instructions when executed further cause the computing device to:determine, based at least in part on the first stereo disparity, a first length between a first right eye and a first left eye of the first face; determine, based at least in part on the second disparity, a second length between a second right eye and a second left eye of the second face; determine at least one of the first length or the second length is greater or less than an acceptable range of eye lengths; and determine at least one of the first face or the second face is not a user based on at least one of the first length or the second length being greater or less than the acceptable range of eye lengths. 18. The computing device of claim 17 , wherein the instructions when executed further cause the computing device to:detect the one or more first bounding boxes from first pixels associated with the first face in the image data; and determining the one or more second bounding boxes associated with the second face in the image data.
意見反饋
使用該功能遇到問題