Unveil the potential of siamese framework for visual tracking

•We propose a global, nonlinear and scale-invariant similarity measuring manner called Dynamic Cross-Attention (DCA) based on the Transformer Encoder.•We redesign the Siamese framework based on DCA to achieve more efficient feature extraction, aggregation and refinement, including introducing more advanced backbones, applying the bidirectional pyramidal neck and modifying head.•The proposed SiamDCA performs favorably compared to other SOTA Siamese and CF trackers on VOT challenges and large-scale benchmarks such as ITB, NFS, LaSOT and GOT-10k.