Text-to-Image-to-Text Translation using Cycle Consistent Adversarial Networks
This addresses the issue of semantic accuracy in text-to-image generation for applications like content creation, but it is incremental as it builds on existing GAN-based methods.
The paper tackles the problem of text-to-image translation where generated images often don't accurately reflect input sentences, by using a captioning network to caption generated images and exploiting the distance between ground truth and generated captions to improve the network, showing extensive comparisons with existing methods.
Text-to-Image translation has been an active area of research in the recent past. The ability for a network to learn the meaning of a sentence and generate an accurate image that depicts the sentence shows ability of the model to think more like humans. Popular methods on text to image translation make use of Generative Adversarial Networks (GANs) to generate high quality images based on text input, but the generated images don't always reflect the meaning of the sentence given to the model as input. We address this issue by using a captioning network to caption on generated images and exploit the distance between ground truth captions and generated captions to improve the network further. We show extensive comparisons between our method and existing methods.