Adversarially trained RTMpose: A high-performance, non-contact method for detecting Genu valgum in adolescents

Shichun He, Meiqi Wei, Deyu Meng, Zongnan Lv, Hongzhi Guo, Guang Yang, Ziheng Wang

Abstract: Highlights•High-performance, non-contact method for Genu valgum detection using adversarial training.•End-to-end Genu valgum prediction model with RTMpose for landmark extraction.•Enhanced model robustness by aligning pose estimation biases with adversarial noise.