A Systematic Review of Text-to-Image and Image-to-Image Synthesis Models
Abstract
The generation of realistic images with machine learning in particular has gained popularity in the disciplines of computer vision and NLP. This study strives to carry out a systematic literature review on image generation techniques placing emphasis on Text-to-Image (T2I) and Image-to-Image (I2I) Synthesis. Such methods can be applied for creating detailed and photo realistic visuals from textual prompts. The survey particularly focuses on recent developments in I2I synthesis models which include SC-StyleGAN, DG Norm, LSCA architecture, Sketch Guided Latent Diffusion Model (SGLDM) and some existing models for T2I synthesis like DE-GAN, PM-GAN and SSTiGAN. Additionally, we explore the idea of Text-Guided Image-To-Image (TG-I2I) Synthesis, a multimodal approach which produces images by using both text and image inputs. We consider several publicly available datasets such as CelebAMask-HQ, MS-COCO, CUB, CUHK Face Sketch, among others, that are necessary for training the models. To evaluate the effectiveness of the surveyed frameworks in image quality, diversity and realism, evaluation metrics like Inception Score (IS), Fréchet Inception Distance (FID) and R-Precision are put forth. In particular, this survey tries to address the methodological gaps in this area and maps out areas for further study, emphasizing the emerging realm of Text-Guided Image Synthesis.
How strong is the evidence?
Evidence Hub reads this as a systematic review. Its evidence level is 1 of 5 (Highest · pooled evidence). Pooled evidence from several studies sits at the top of the evidence pyramid.
Read the paper
Related studies
- Learning from First-Generation Qualitative Approaches in the IS Discipline: An Evolutionary View and Some Implications for Authors and Evaluators (PART 1/2)
- Commentary on the clinical and preclinical dosage limits of interstitially administered magnetic fluids for therapeutic hyperthermia based on current practice and efficacy models
- Prosthesis: New Technological Opportunities and Innovative Biomedical Devices
- The history and future of neural modeling for cochlear implants
- Ryanodine receptors: waking up from refractoriness
- Novel Transgenic Rabbit Model Sheds Light on the Puzzling Role of Matrix Metalloproteinase-12 in Atherosclerosis
Go deeper in Evidence Hub
Evidence Hub adds a study data sheet (design, sample size, evidence level, key numbers), an AI summary and a chat grounded in the paper's text.