Multimodal Multi-Encoder

AbstractAn image caption should fluently present the essential information in a given image, including informative, fine-grained entity mentions and the manner in which these entities interact. However, current captioning models …...