Research map / Image generation / GANs
DF-GAN: A Simple and Effective Baseline for Text-to-Image Synthesis
Check 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 |
|---|---|---|---|
| Replace the stacked backbone with a one-stage backbone (ablation) | Text-to-image GAN · CUB test | FID: 51.34 -> 43.45 | Helped |
| Add Matching-Aware Gradient Penalty (ablation) | OS-B · CUB test | SC: 1.46 -> 3.55 | Helped |
| Replace DAMSM supervision with the Target-Aware Discriminator using MA-GP and One-Way Output (ablation) | OS-B · CUB test | SC: 1.79 -> 4.61 | Helped |
| Add One-Way Output (ablation) | OS-B with MA-GP · CUB test | FID: 32.52 -> 23.16 | Helped |
| Replace concatenation with Conditional Batch Normalization for text-image fusion (ablation) | MA-GP GAN · CUB test | IS: 4.57 -> 4.81 | Helped |
| Replace concatenation with Adaptive Instance Normalization for text-image fusion (ablation) | MA-GP GAN · CUB test | IS: 4.57 -> 4.85 | Helped |
| Replace one Affine Transformation layer with deep text-image fusion blocks (ablation) | MA-GP GAN · CUB test | FID: 17.43 -> 14.81 | Helped |
| Use the Target-Aware Discriminator instead of a fixed DAMSM extra network (ablation) | OS-B · CUB test | FID: 36.72 -> 23.16 | Helped |
| Use deep text-image fusion blocks rather than concatenation (ablation) | MA-GP GAN · CUB test | IS: 4.57 -> 5.10 | Helped |
Closest papers by meaning
- DM-GAN: Dynamic Memory Generative Adversarial Networks for Text-To-Image Synthesis (CVPR 2019 · similarity 0.92)
- StyleGAN-T: Unlocking the Power of GANs for Fast Large-Scale Text-to-Image Synthesis (ICML 2023 · similarity 0.91)
- Adversarial Text to Continuous Image Generation (CVPR 2024 · similarity 0.89)
- Exploring Sparse MoE in GANs for Text-conditioned Image Synthesis (CVPR 2025 · similarity 0.88)
- You Only Sample Once: Taming One-Step Text-to-Image Synthesis by Self-Cooperative Diffusion GANs (ICLR 2025 · similarity 0.88)
- E$^2$GAN: Efficient Training of Efficient GANs for Image-to-Image Translation (ICML 2024 · similarity 0.88)
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.