Multiple Context Learning Networks for Visual Question Answering

Figure 2

Overall flowchart of the MCLN that consists of three subnetworks. (a) Image and question representation; (b) multiple context learning; (c) multimodal fusion and answer prediction.