Object detect, segment, and pose using one tracking framework
Object detect, segment, and pose using one tracking framework
Do Different Tracking Tasks Require Different Appearance Models?
arXiv paper abstract https://arxiv.org/abs/2107.02156
arXiv PDf paper https://arxiv.org/pdf/2107.02156.pdf
GitHub https://github.com/Zhongdao/UniTrack
Tracking objects … novel approaches proposed by the community are usually specialised to fit only one specific setup.
… present UniTrack, a unified tracking solution to address five different tasks within the same framework.
UniTrack consists of a single and task-agnostic appearance model, which can be learned in a supervised or self-supervised fashion, and multiple “heads” to address individual tasks and that do not require training.
… most tracking tasks can be solved within this framework, and that the same appearance model can be used to obtain performance that is competitive against specialised methods for all the five tasks considered.
… allows us to analyse appearance models obtained with the most recent self-supervised methods, thus significantly extending their evaluation and comparison to a larger variety of important problems. …
Stay up to date. Subscribe to my posts https://morrislee1234.wixsite.com/website/contact
Web site with my other posts by category https://morrislee1234.wixsite.com/website
LinkedIn https://www.linkedin.com/in/morris-lee-47877b7b
