High-Speed Tracking with Kernelized Correlation Filters
University of Coimbra · Institute for Systems Engineering and Computers
Abstract
The core component of most modern trackers is a discriminative classifier, tasked with distinguishing between the target and the surrounding environment. To cope with natural image changes, this classifier is typically trained with translated and scaled sample patches. Such sets of samples are riddled with redundancies-any overlapping pixels are constrained to be the same. Based on this simple observation, we propose an analytic model for datasets of thousands of translated patches. By showing that the resulting data matrix is circulant, we can diagonalize it with the discrete Fourier transform, reducing both storage and computation by several orders of magnitude. Interestingly, for linear regression our…
Citation impact
- FWCI
- 200.92
- Percentile
- 100%
- References
- 48
Authors
4- JFJoão F. HenriquesCorresponding
University of Coimbra, Institute for Systems Engineering and Computers
- RCRui Caseiro
University of Coimbra, Institute for Systems Engineering and Computers
- PMPedro Martins
Institute for Systems Engineering and Computers, University of Coimbra
- JBJorge Batista
University of Coimbra, Institute for Systems Engineering and Computers
Topics & keywords
- Circulant matrix
- Artificial intelligence
- Computer science
- BitTorrent tracker
- Kernel (algebra)
- Discriminative model
- Pattern recognition (psychology)
- Mathematics
- Reduced inequalities