:: Jan Prokaj :::...
Using 3D scene structure to improve tracking
Aerial imagery collected over urban areas contains many occlusions for vehicles
Current tracking algorithms cannot track vehicles through such occlusions due to weak object appearance and complexity of motion prediction over long period
Use known 3D scene structure to estimate a dynamic occlusion map: binary map, which indicates what regions of the image are occluders of moving objects; dynamic due to camera motion
Cluster track initiations and terminations near an occlusion map to find sources and sinks
Find corresponding source-sink pairs using the Hungarian algorithm
of tracks at the sink with a
of tracks at the source using dynamic programming
Merge the matched tracks
Database of 3D models
(93 models, 40.5 MiB)
Sequences used in experiments
Updated occlusion maps
Information coming soon!
J. Prokaj and
. Using 3D Scene Structure to Improve Tracking. In
IEEE Conference on Computer Vision and Pattern Recognition
, pages 1337-1344, 2011.