Abstract: Recent transformer-based visual tracking models have demonstrated superior performance. However, prior works have been resource-intensive, requiring massive GPU training hours. This resource ...