Tracking
Description:
Tracking is the problem of following multiple moving objects over several time steps. Classic applications include tracking people or vehicles in videos and tracking cells in time series of microscopy images in biology. We explore formulations tailored to the specific application, their theoretical properties, and algorithms that efficiently solve these problems. Embedded in neural networks, we have achieved high-quality empirical results.
Literatur:
- "LMGP: Lifted Multicut Meets Geometry Projections for Multi-Camera Multi-Object Tracking", Nguyen, Duy M. H. and Henschel, Roberto and Rosenhahn, Bodo and Sonntag, Daniel and Swoboda, Paul, Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition (CVPR) 2022
- "Making Higher Order MOT Scalable: An Efficient Approximate Solver for Lifted Disjoint Paths", Hornakova, Andrea and Kaiser, Timo and Swoboda, Paul and Rolinek, Michal and Rosenhahn, Bodo and Henschel, Roberto, Proceedings of the IEEE International Conference on Computer Vision (ICCV) 2021
- "Lifted Disjoint Paths with Application in Multiple Object Tracking", Hornakova, Andrea and Henschel, Roberto and Rosenhahn, Bodo and Swoboda, Paul, Proceedings of the International Conference on Machine Learning (ICML) 2020
- "A Primal-Dual Solver for Large-Scale Tracking-by-Assignment", Haller, Stefan and Prakash, Mangal and Hutschenreiter, Lisa and Pietzsch, Tobias and Rother, Carsten and Jug, Florian and Swoboda, Paul and Savchynskyy, Bogdan, Proceedings of the International Conference on Artificial Intelligence and Statistics (AISTATS) 2020