Creating Artificial Birds: A Novel Neural Net Generates Realistic Pictures of Birds from Text Using Common Sense
Summary: Researchers from China have developed a neural network which generates high quality bird images using textual descriptions. They use common-sense to enhance the generated image with three different resolution levels, and achieve competitive scores with other methods of neural networks. The network was created using a generative antagonistic network, and trained on a dataset that included bird images and descriptions. Its goal is to promote the development of text-to image synthesis.
Source: Intelligent Computing.
A group of Chinese researchers built a generative network using data that represents common sense knowledge in an attempt to create high-quality images from text descriptions.