Viewing results

View3D

View3D offers the same interactive functionality as Label3D without some labeling and saving features. It makes it easy to view 3D keypoint predictions over multi-view images without the labeling-specific features of Label3D. First just construct a View3D instance as you would for Label3D then load the 3D keypoints with the loadFrom3D method.

viewGui = View3D(params, videos, skeleton);

% Load dannce predictions from a file. pts3d.pred is nFrames x 3 x nKeypoints 
pts3d = load('/path/to/dannce/predictions.mat')
viewGui.loadFrom3D(pts3d.pred) 

Making videos

Label3D and View3D both can create videos of your labels/predictions using Animator's writeVideo method. For instance, to make a video of the keypoint predictions from the example above:

framesInVideo = 1:1000
savePath = 'video.mp4'
viewGui.writeVideo(framesInVideo, savePath, 'FPS', 30, 'Quality', 50)

writeVideo can produce videos in .mp4, .avi, and .gif formats and more. For more information look at help Animator.writeVideo and help write_frames. Please note that .mp4 video writing is not supported in Linux distributions.