Stacking VAE and GAN for Context-aware Text-to-Image Generation | IEEE Conference Publication | IEEE Xplore