论文信息 - Text to Complicated Image Synthesis with Segmentation Information Guidance

Text to Complicated Image Synthesis with Segmentation Information Guidance

In this paper, we propose a novel method called Segmentation Information Guidance (SIG). In this method, additional segmentation information is added to guide the process of text to complicated image synthesis. We demonstrate the effectiveness of SIG model on Microsoft Common Objects in Common (MSCOCO) dataset. It proves that the image results generated by directly using the segmentation image are more authentic and coherent than that without background.

[1] Bernt Schiele,et al. Generative Adversarial Text to Image Synthesis , 2016, ICML.

[2] Jürgen Schmidhuber,et al. Long Short-Term Memory , 1997, Neural Computation.

[3] Geoffrey E. Hinton,et al. ImageNet classification with deep convolutional neural networks , 2012, Commun. ACM.

[4] Sanja Fidler,et al. Skip-Thought Vectors , 2015, NIPS.

[5] Soumith Chintala,et al. Unsupervised Representation Learning with Deep Convolutional Generative Adversarial Networks , 2015, ICLR.

[6] Dimitris N. Metaxas,et al. StackGAN: Text to Photo-Realistic Image Synthesis with Stacked Generative Adversarial Networks , 2016, 2017 IEEE International Conference on Computer Vision (ICCV).

[7] Jimmy Ba,et al. Adam: A Method for Stochastic Optimization , 2014, ICLR.

[8] Yoshua Bengio,et al. Generative Adversarial Nets , 2014, NIPS.

[9] Pietro Perona,et al. Microsoft COCO: Common Objects in Context , 2014, ECCV.

[10] Zhe Gan,et al. AttnGAN: Fine-Grained Text to Image Generation with Attentional Generative Adversarial Networks , 2017, 2018 IEEE/CVF Conference on Computer Vision and Pattern Recognition.

[11] Yoshua Bengio,et al. Empirical Evaluation of Gated Recurrent Neural Networks on Sequence Modeling , 2014, ArXiv.