Fig. 2 | Scientific Reports

Fig. 2

From: Construction of intelligent evaluation model for physical education classroom in primary and secondary schools based on posture estimation and motion recognition

Fig. 2

Schematic diagram of the Pedagogical Alignment Modeling. Student motion input \(x_t^{(i)}\) is encoded by Bi-GRU into \(h_t^{(i)}\), while the teacher instruction sequence \(\theta _{1:T}\) is encoded into \(\tau _t\) via a transformer. Cosine similarity is computed between \(h_t^{(i)}\) and \(\tau _t\), with delay-aware alignment across a trailing window. The final conformity vector \(\pi ^{(i)}\) is computed by attention-weighted integration of the alignment scores.

Back to article page