Research map / Image generation / Text-to-image diffusion
Photographic Text-to-Image Synthesis With a Hierarchically-Nested Adversarial Network
Read the paperCheck your idea against 108,967 papers
AI-extracted summary. Written by AI from the paper's title only (no summary yet). Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.
Measured effects reported in the paper
| What was changed | Where | Measured | Effect |
|---|---|---|---|
| use hierarchically-nested discriminators at 64, 128, and 256 scales (ablation) | HDGAN · CUB | Inception score: 3.52{\pm}.04 -> 4.15{\pm}.05 | Helped |
| add the 64-scale discriminator to the 128- and 256-scale discriminators (ablation) | HDGAN · CUB | Inception score: 3.99{\pm}.04 -> 4.15{\pm}.05 | Helped |
| add the 128-scale discriminator to the 64- and 256-scale discriminators (ablation) | HDGAN · CUB | Inception score: 4.14{\pm}.03 -> 4.15{\pm}.05 | No clear effect |
| add the 128-scale discriminator to the 64- and 256-scale discriminators (ablation) | HDGAN · COCO | Inception score: 11.29{\pm}.18 -> 11.86{\pm}.18 | Helped |
| add the local image loss (ablation) | HDGAN · Oxford-102 | Inception score: 3.12{\pm}.02 -> 3.45{\pm}.07 | Helped |
| add the local image loss (ablation) | HDGAN · Oxford-102 | VS similarity: .263{\pm}.130 -> .296{\pm}.130 | Helped |
| share the top layers of the hierarchical discriminators instead of using independent discriminators (ablation) | HDGAN · text-to-image synthesis | performance | No clear effect |
| use instance normalization or layer normalization instead of the reported normalization (ablation) | HDGAN · text-to-image synthesis | performance | Hurt |
| use hierarchical-nested adversarial training rather than StackGAN's intermediate text conditioning | HDGAN · CUB | Inception score: 3.7 -> 3.45 | Helped |
Closest papers by meaning
- Adversarial Learning of Semantic Relevance in Text to Image Synthesis (AAAI 2019 · similarity 0.92)
- Perceptual Pyramid Adversarial Networks for Text-to-Image Synthesis (AAAI 2019 · similarity 0.90)
- Text to Image Generation With Semantic-Spatial Aware GAN (CVPR 2022 · similarity 0.89)
- DF-GAN: A Simple and Effective Baseline for Text-to-Image Synthesis (CVPR 2022 · similarity 0.89)
- Adversarial Text to Continuous Image Generation (CVPR 2024 · similarity 0.87)
- Unlocking Pre-trained Image Backbones for Semantic Image Synthesis (CVPR 2024 · similarity 0.87)
Writing a paper in this area? scime draft is a free LaTeX editor with AI included, real-time co-authors and a check of your references and venue rules before you submit. It also imports OpenAI Prism projects.