Research map / Image generation / Text-to-image diffusion
Text to Image Generation With Semantic-Spatial Aware GAN
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 DF-GAN UPBlocks with SSACN blocks, without DAMSM (ablation) | DF-GAN · CUB test | IS: 4.86 $\pm$ 0.04 -> 4.97 $\pm$ 0.09 | Helped |
| replace DF-GAN UPBlocks with SSACN blocks, without DAMSM (ablation) | DF-GAN · CUB test | FID: 19.24 -> 18.54 | Helped |
| add DAMSM (ablation) | SSA-GAN with SSACN · CUB test | IS: 4.97 $\pm$ 0.09 -> 5.07 $\pm$ 0.04 | Helped |
| add DAMSM (ablation) | SSA-GAN with SSACN · CUB test | FID: 18.54 -> 15.61 | Helped |
| fine-tune the text encoder jointly with the generator (ablation) | SSA-GAN with SSACN and DAMSM · CUB test | IS: 5.07 $\pm$ 0.04 -> 5.17 $\pm$ 0.08 | Helped |
| fine-tune the text encoder jointly with the generator (ablation) | SSA-GAN with SSACN and DAMSM · CUB test | FID: 15.61 -> 16.58 | Hurt |
| increase the number of mask stages from 2 to 4 (ablation) | SSA-GAN · CUB test | IS: 4.98 $\pm$ 0.09 -> 5.05 $\pm$ 0.05 | Helped |
| increase the number of mask stages from 2 to 4 (ablation) | SSA-GAN · CUB test | FID: 19.69 -> 15.03 | Helped |
| increase the number of mask stages from 4 to 6 (ablation) | SSA-GAN · CUB test | IS: 5.05 $\pm$ 0.05 -> 4.97 $\pm$ 0.04 | Mixed |
| increase the number of mask stages from 4 to 6 (ablation) | SSA-GAN · CUB test | FID: 15.03 -> 16.62 | Mixed |
| increase the number of mask stages from 6 to 7 (ablation) | SSA-GAN · CUB test | IS: 4.97 $\pm$ 0.04 -> 5.07 $\pm$ 0.04 | Helped |
| increase the number of mask stages from 6 to 7 (ablation) | SSA-GAN · CUB test | FID: 16.62 -> 15.61 | Helped |
Closest papers by meaning
- DF-GAN: A Simple and Effective Baseline for Text-to-Image Synthesis (CVPR 2022 · similarity 0.91)
- Adversarial Learning of Semantic Relevance in Text to Image Synthesis (AAAI 2019 · similarity 0.90)
- Generative Adversarial Text to Image Synthesis (ICML 2016 · similarity 0.90)
- Adversarial Text to Continuous Image Generation (CVPR 2024 · similarity 0.89)
- Unlocking Pre-trained Image Backbones for Semantic Image Synthesis (CVPR 2024 · similarity 0.89)
- MARS: Mixture of Auto-Regressive Models for Fine-grained Text-to-image Synthesis (AAAI 2025 · 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.