Proposed sliding-window based target pose estimation and prediction method achieving 91% accuracy

Developed three-stage recognition method preventing information loss during downsampling, improving long-range target pose estimation accuracy by 16%

Code Repository: https://github.com/HUSTLYRM/HUST_HeroAim_2024

Won the Excellence Award in the RoboMaster Open Source Award, receiving a cash prize of 3000 RMB.