wangxiao5791509 / TNL2K_evaluation_toolkit

Towards More Flexible and Accurate Object Tracking with Natural Language: Algorithms and Benchmark (CVPR 2021)
https://sites.google.com/view/langtrackbenchmark/
43 stars 7 forks source link

Results reported in the paper #15

Open npmhung opened 2 years ago

npmhung commented 2 years ago

Hello,

I would like to ask if all the methods reported in the papers have been tuned on the TNL2K training set?

Best

wangxiao5791509 commented 2 years ago

@npmhung Hi, part of the evaluated language-based trackers are trained using the training subset of TNL2K, but the bbox-based trackers are tested using the pre-trained models released by corresponding authors.