Research map / Image generation / Text-to-image diffusion
Diffusion Probe: Generated Image Result Prediction Using CNN Probes
Bukun Huang, Benlei Cui, Zhizeng Ye, Xuemei Dong, Tuo Chen, Hui Xue, Dingkang Yang, Longtao Huang, Haiwen Hong, Jingqun Tang
Read the paperCheck your idea against 137,217 papers
AI-extracted. The measured effects below were read by AI from the paper's full text (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 |
|---|---|---|---|
| extract attention at step 5 instead of step 1 (ablation) | Diffusion Probe on FLUX · MSCOCO test | SRCC: 0.52 -> 0.76 | Improved |
| extract attention at step 10 instead of step 5 (ablation) | Diffusion Probe on FLUX · MSCOCO test | SRCC: 0.76 -> 0.79 | Improved |
| extract attention at step 15 instead of step 10 (ablation) | Diffusion Probe on FLUX · MSCOCO test | SRCC: 0.79 -> 0.78 | No clear effect |
| gate prompts with probe before LLM refinement (ablation) | FLUX.1-dev · MSCOCO prompt optimization | ImageReward: 1.02 -> 1.10 | Improved |
| always use LLM refinement instead of probe gating | FLUX.1-dev · MSCOCO prompt optimization | CLIP Score: 32.85 -> 33.11 | Improved |
| probe-based seed selection over random seed (ablation) | SDXL · MSCOCO seed selection | ImageReward: 0.71 -> 0.82 | Improved |
| probe-based seed selection over random seed | FLUX.1-dev · MSCOCO seed selection | CLIP Score: 31.37 -> 31.41 | No clear effect |
| use probe to enrich Flow-GRPO batches with high-quality samples | Flow-GRPO on FLUX.1-dev · PickScore training curve | convergence: 2.5x more high-quality samples | Improved |
Closest papers by meaning
- Faster Diffusion: Rethinking the Role of the Encoder for Diffusion Model Inference (NeurIPS 2024 · similarity 0.90)
- Elucidating the design space of classifier-guided diffusion generation (ICLR 2024 · similarity 0.90)
- InitNO: Boosting Text-to-Image Diffusion Models via Initial Noise Optimization (CVPR 2024 · similarity 0.90)
- Schedule On the Fly: Diffusion Time Prediction for Faster and Better Image Generation (CVPR 2025 · similarity 0.89)
- Lotus: Diffusion-based Visual Foundation Model for High-quality Dense Prediction (ICLR 2025 · similarity 0.89)
- SwiftBrush: One-Step Text-to-Image Diffusion Model with Variational Score Distillation (CVPR 2024 · similarity 0.89)
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.
