Fig. 4.

C-Net architecture. In the figure, $7\times 7$conv and $3\times 3$conv represent $7\times 7$ and $3\times 3$ convolution operation, respectively, while $\times 2$up and $\times 4$up represent bilinear up-sampling operation with 2 and 4 scale factors, respectively. The numbers on the arrows represent the sizes of feature maps. The pyramid parsing module (PPM) [21] was used to extract multi-scale features.