Research map / Image generation / Text-to-image diffusion
Controllable Text-to-Image Generation
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 ControlGAN instead of AttnGAN | text-to-image generation · CUB | Inception Score: 4.36 $\pm$ .03 -> 4.58 $\pm$ .09 | Helped |
| Use ControlGAN instead of AttnGAN | text-to-image generation · CUB | Top-1 Acc(%): 67.82 $\pm$ 4.43 -> 69.33 $\pm$ 3.23 | Helped |
| Use ControlGAN instead of AttnGAN | text-to-image generation · CUB | L2 reconstruction error: 0.26 -> 0.18 | Helped |
| Use ControlGAN instead of AttnGAN | text-to-image generation · COCO | Inception Score: 25.89 $\pm$ .47 -> 24.06 $\pm$ .60 | Hurt |
| Use ControlGAN instead of AttnGAN | text-to-image generation · COCO | Top-1 Acc(%): 85.47 $\pm$ 3.69 -> 82.43 $\pm$ 2.43 | Hurt |
| Use ControlGAN instead of AttnGAN | text-to-image generation · COCO | L2 reconstruction error: 0.40 -> 0.17 | Helped |
| Add channel-wise attention alongside spatial attention (ablation) | ControlGAN generator · CUB attribute manipulation | controllability | Helped |
| Add the word-level discriminator (ablation) | ControlGAN · CUB attribute manipulation | preservation of unrelated image content | Helped |
| Replace the word-level discriminator with a text-adaptive discriminator using global pooling (ablation) | ControlGAN · CUB attribute manipulation | preservation of unrelated image content | Hurt |
| Add perceptual loss (ablation) | ControlGAN · text-modified image generation | preservation of content related to unmodified text | Helped |
Closest papers by meaning
- Text-Adaptive Generative Adversarial Networks: Manipulating Images with Natural Language (NeurIPS 2018 · similarity 0.90)
- Lightweight Generative Adversarial Networks for Text-Guided Image Manipulation (NeurIPS 2020 · similarity 0.89)
- Adversarial Text to Continuous Image Generation (CVPR 2024 · similarity 0.88)
- Compositional Text-to-Image Synthesis with Attention Map Control of Diffusion Models (AAAI 2024 · similarity 0.86)
- InstantPainting: Expanding GANs for Efficient Text-Conditioned Image Generation Platform (AAAI 2025 · similarity 0.86)
- How Control Information Influences Multilingual Text Image Generation and Editing? (NeurIPS 2024 · similarity 0.85)
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.