Research on Video Captioning Based on Multifeature Fusion

Figure 5

Three-dimensional convolution and two-stream expansion 3D convolution network structure. (a) 3D Incception V1. (b) I3D.

(a)

(b)