Spatio-temporal mix deformable feature extractor in visual tracking

Yucheng Huang, Ziwang Xiao, Eksan Firkat, Jinlai Zhang, Danfeng Wu, Askar Hamdulla

Abstract: Highlights•Multi-frame fusion module improved upon ACMix.•Spatio-temporal context driven deformation for improving multi-frame trackers.•Spatio-temporal transformation for improving synthesized data.•Sufficient experiments have demonstrated its effectiveness.