FFT-SCSHO-SVM
International Journal of e-Collaboration · 20.10.2026 · 22 (1) · 1-28
DOI: 10.4018/ijec.425348 ↗Kurz gefasst
Fachartikel in der Zeitschrift International Journal of e-Collaboration, erschienen am 20.10.2026. Verfasst von Lijuan Jia. Die Arbeit nennt 35 Literaturangaben. Der Volltext ist frei zugänglich (Open Access, CC BY 3.0).
Abstract (Originalsprache)
Real-time basketball free throw detection faces interference from varying lighting, viewpoints, and complex motions. This work presents an FFT-SCSHO-SVM hybrid model integrating FFT frequency-domain pose encoding and SCSHO optimization for feature selection and SVM hyperparameter tuning. Pose time sequences are converted into frequency spectra representing shooting rhythms, while SCSHO boosts model discrimination and stability. Tests on BAD and self-built CFD under diverse lighting and occlusions yield a peak accuracy of 98.3%, outperforming four mainstream benchmarks by 2.7–6.1%. Cross-dataset validation shows slight symmetric accuracy losses, and Gaussian noise tests retain over 94% accuracy. Its CPU inference speed of 54.3 fps satisfies real-time demands. This interpretable, efficient model facilitates real-time sports analysis and intelligent coaching.
Zitieren
APA
Jia, L. (2026). FFT-SCSHO-SVM. International Journal of e-Collaboration, 22(1), 1-28. https://doi.org/10.4018/ijec.425348
Vancouver
Jia L. FFT-SCSHO-SVM. International Journal of e-Collaboration. 2026;22(1):1-28. doi:10.4018/ijec.425348
BibTeX
@article{Jia2026,
author = {Jia, Lijuan},
title = {FFT-SCSHO-SVM},
journal = {International Journal of e-Collaboration},
year = {2026},
volume = {22},
number = {1},
pages = {1--28},
doi = {10.4018/ijec.425348}
}
EUVIMED