A Text-to-Image Generation Approach Combining Mutual Information Maximization
In this paper,we propose a text-to-image generation method incorporating mutual information maximization,which aims to learn the mutual information between text and image and maximize the mutual information between the two.The results of CUB tests show that the algorithm proposed in this paper can improve the generated sample discrepancies to a greater extent and has higher accuracy in objective assessment,thus more closely resembling real natural images.
mutual information maximizationtext transformationimage generationnatural imagesartificial intelligence
万方数据
