Our research focuses on developing a shot-by-shot performance evaluati
on system for smart courts. This system utilizes advanced stereoscopic im
aging and computer vision technologies to capture and analyze players'
shot movements in singles matches, providing scientific performance dat
a. It consists of four main components: a stereoscopic imaging system, 2
D/3D trajectory calculation, semantic analysis, and performance metrics c
alculation. The system aims to enhance the user experience, improve spor
ts learning efficiency, and increase the entertainment value of badminton.
The stereoscopic imaging system employs synchronized high-speed came
ras to capture gameplay from multiple angles, allowing for comprehensiv
e analysis of the badminton trajectory. The 2D/3D trajectory calculation c
omponent uses deep learning techniques to track the shuttlecock's traje
ctory and calculate its 3D trajectory, enabling accurate measurement of p
osition and movement from various perspectives. The semantic analysis s
egment detects shot events, segments the trajectory, and eliminates devia
tions caused by detection errors, providing precise information about key
shuttlecock events. The performance metrics calculation component extra
cts data from the 3D trajectory, including shuttlecock speed, shot angle, a
nd ball types, offering essential reference information for players and coac
hes to analyze and evaluate performance.
By providing accurate and comprehensive records of badminton shot perf
ormance, this system assists players and coaches in understanding their s

kills, identifying strengths, and areas for improvement. It also enhances th
e viewing experience for spectators and fans, deepening their understandi
ng of the sport. The system's applications extend beyond smart badmint
on venues to training and competition venues, ultimately elevating the qu
ality and competitiveness of badminton sports.