Skip to main content
. 2024 Oct 28;7:26. doi: 10.1186/s42492-024-00176-5

Fig. 2.

Fig. 2

Pipeline of the framework. The network is divided into three components: motion encoding, sequence predictor, and motion decoding. The sequence predictor consists of a shortcut connection, a discrete cosine transform (DCT), spatial-frequency block (SF block), and spatial-temporal block (ST block)