[Retracted] Human Posture Recognition and Estimation Method Based on 3D Multiview Basketball Sports Dataset

Table 1

Structure of feature extraction network.

Module	Structure	Size	Number of output channels	Proportion to original image
Convolutional layer 1	Conv2D	3 × 3	64	1
Convolutional layer 2	Maxpooling	2 × 2	64	1/2
Convolutional layer 3	Comv2D	3 × 3	128	1/2
Convolutional layer 4	Maxpooling	2 × 2	128	1/4
Convolutional layer 5	Conv2D	3 × 3	256	1/4
Convolutional layer 6	Conv2D	3 × 3	256	1/4
Fully connected layer 1	Dense	—	256	—
Fully connected layer 2	Dense	—	512	—