Markerless motion capture
A system for markerless motion capture may include an image capture apparatus that obtains video including live movement of one or more subjects. An estimation module analyzes frames of the video and estimates the location of a plurality of keypoints on each of the subjects. A propagation module applies the keypoints corresponding to a first subject to the first subject additional frames of the video. The propagation module does the same for additional subjects in the video. A conversion module converts the plurality of keypoints from each frame of the video into a digital media file having a first file format and then converts the digital media file in the first file format into a digital media file having second file format. The system also includes a data storage apparatus configured to save the digital media files in the first and second file formats.
1 . A system for markerless motion capture comprising:
an image capture apparatus configured to obtain video including live movement of a first subject, wherein the video includes a plurality of sequential frames;
a bounding module configured to analyze one or more frames of the video and identify the first subject in one or more frames of the video and apply a bounding box around the first subject in each of the one or more frames of the video;
an estimation module configured to analyze each of the one or more frames of the video having a bounding box around the first subject and estimate a location for each of a plurality of keypoints on the first subject;
a conversion module configured to convert the estimated location for the plurality of keypoints from each frame of the video containing at least one keypoint into a digital media file having a first file format;
the conversion module further configured to convert the digital media file in the first file format into a digital media file having a second file format; and
a data storage apparatus configured to save the digital media file in the first file format and save the digital media file in the second file format.
2 . The system of claim 1 , wherein the image capture apparatus is a camera configured to capture live movement of the first subject.
3 . The system of claim 1 , wherein the first file format is a biovision hierarchy (BVH) format.
4 . The system of claim 1 , wherein the second file format is a Filmbox (FBX) format.
5 . The system of claim 1 , further comprising a pattern module configured to analyze a plurality of keypoints in a plurality of sequential frames of the video and identify a movement pattern comprising a relative position of a plurality of keypoints in a plurality of sequential frames.
6 . The system of claim 5 , wherein the image capture apparatus further configured to obtain video including live movement of a plurality of subjects.
7 . The system of claim 6 , wherein the movement pattern comprises the relative position of a first plurality of keypoints corresponding to a first subject in a plurality of sequential frames and the relative position of second plurality of keypoints corresponding to a second subject in the plurality of sequential frames.
8 . The system of claim 1 , wherein the conversion module is further configured to convert the estimated location for the plurality of keypoints from each frame of the video containing at least one keypoint into the digital media file having the second file format.
9 . The system of claim 1 , wherein the conversion module is further configured to improve smoothness of motion in the digital media file having the first file format by converting the digital media file having the first file format into a reconverted plurality of keypoints and converting the reconverted plurality of keypoints into an improved digital media file having the first file format.
10 . The system of claim 1 , wherein the plurality of keypoints on the first subject represent the first subject in three dimensional space.
11 . A method for converting markerless motion capture comprising:
obtaining video including live movement of a first subject, wherein the video includes a plurality of sequential frames;
identifying the first subject in one or more frames of the video;
applying a bounding box around the first subject in each of the one or more frames of the video;
estimating a location for each of a plurality of keypoints on the first subject in each bounding box around the first subject in each of the one or more frames of the video;
converting the estimated location for the plurality of keypoints from each frame of the video containing at least one keypoint into a digital media file having a first file format;
converting the digital media file in the first file format into a digital media file having a second file format; and
storing the digital media file in the first file format and storing the digital media file in the second file format.
12 . The method of claim 11 , wherein the video is obtained from a live video feed.
13 . The method of claim 11 , wherein the video is prerecorded and obtained from data storage.
14 . The method of claim 11 , wherein the first file format is a biovision hierarchy (BVH) format.
15 . The method of claim 11 , wherein the second file format is a Filmbox (FBX) format.
16 . The method of claim 11 , further comprising analyzing a plurality of keypoints in a plurality of sequential frames of the video to identify a movement pattern, wherein the movement pattern comprises relative positions of a plurality of keypoints in a plurality of sequential frames.
17 . The method of claim 16 , wherein the movement pattern comprises the relative position of a first plurality of keypoints corresponding to a first subject in a plurality of sequential frames and the relative position of second plurality of keypoints corresponding to a second subject in the plurality of sequential frames.
18 . The method of claim 11 , further comprising, prior to converting the estimated location for the plurality of keypoints to the first file format and prior to converting the first file format to the second file format, adjusting a location of a first keypoint in a first frame relative to a location of the first keypoint in a second frame that is adjacent to the first frame such that a distance between the first keypoint in the first frame and a second keypoint in the first frame is substantially equal to a distance between the first keypoint in the second frame and the second keypoint in the second frame.
19 . The method of claim 11 , further comprising converting the digital media file having the first file format into a reconverted plurality of keypoints and converting the reconverted plurality of keypoints into an improved digital media file having the first file format.
20 . The method of claim 11 , wherein the plurality of keypoints on the first subject represent the first subject in three dimensional space.