
Experimental Data
This video and tracking data is freely available for research and educational use provided the BORG Lab at Georgia Tech is credited appropriately.
Laser Scans Used For Tracking people
The rest of our laser tracking data can be found on the Laser Tracking webpage.
500 frames of five people walking around the lab, under instruction to not bump into each other. Movies of data can be found here.
3100 frames of five people walking around the lab, under instruction to not bump into each other. Movies of data can be found here.
Data from MCMC Data Association and Sparse Factorization Updating for Real Time Multitarget Tracking with Merged and Multiple Measurements Z. Khan, T. Balch, and F. Dellaert. Submitted to IEEE Transactions on Pattern Analysis and Machine Intelligence, September 2005.
- Download the data set from the FTP site.
- Each of the raw videos (except for files titled _low.mov) use the Quasar Codec which is supported under linux by the libdv library. The Linux Video project provides a handy viewer for these video files called Kino.
- gtvideo.mov is the raw video for the interacting ants sequence. gtvideo_low.mov is the compressed video underlay. The tracking data is stored in gtvideoBTF.tar.gz. You need to directory our trajectory viewer program, TeamView to load the BTF directory which has the ground truth trajectory data in gtvideoBTF.tar.gz and use gtvideo_low.mov as the video underlay.
- The laser data for all 4 lasers tracking 5 people who are instructed to bump into each other is is stored in LaserBumpData.tar.gz. The data for a single laser, laser ID 0, is stored in the BTF file SingleLaserBumpData.tar.gz. You can view the laser data using our trajectory creation and view program, TeamView.
- The videos o[1-16].mov include each of the occlusion sequences we used to test the tracking algorithm. o[1-16]_low.mov have the corresponding lossy compressed video underlays for the trajectories generated by our algorithm. The trajectories are located in the OcclusionTests.tar.gz file. The trajectory BTF file directories can be viewed using our trajectory viewing, editing, and creation program TeamView.
Data from HoneyBee Dance Data Sequences, Experimental Results and Software for *Learning and Inferring Motion Patterns using Parametric Segmental Switching Linear Dynamic Systems*, in International Journal of Computer Vision (IJCV) Special Issue on Learning for Vision (2007).