Abstract: Most state-of-the-art trackers adopt one-stream paradigm, using a single Vision Transformer for joint feature extraction and relation modeling of template and search region images. However, ...
Some results have been hidden because they may be inaccessible to you
Show inaccessible results