Research map / Image generation / Fast diffusion sampling
JoDiffusion: Jointly Diffusing Image with Pixel-Level Annotations for Semantic Segmentation Promotion
Haoyu Wang, Lei Zhang, Wenrui Liu, Dengyang Jiang, Wei Wei, Chen Ding
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 |
|---|---|---|---|
| add mask optimization (tau=20) (ablation) | JoDiffusion · Pascal VOC (synthetic only) | mIoU: 71.37 -> 72.47 | Improved |
| raise mask optimization threshold tau from 20 to 100 (ablation) | JoDiffusion · Pascal VOC (synthetic only) | mIoU: 72.47 -> 72.38 | No clear effect |
| increase generated data from 5k to 40k (ablation) | JoDiffusion · Pascal VOC (synthetic only) | mIoU: 68.54 -> 72.47 | Improved |
| use JoDiffusion instead of Dataset Diffusion (synthetic only) | DeepLabV3 ResNet50 · Pascal VOC | mIoU: 61.6 -> 72.5 | Improved |
| use JoDiffusion instead of Dataset Diffusion (synthetic only) | DeepLabV3 ResNet50 · MS-COCO | mIoU: 32.4 -> 42.6 | Improved |
| add JoDiffusion synthetic data to real data | DeepLabV3 ResNet50 · Pascal VOC | mIoU: 77.4 -> 78.3 | Improved |
| add JoDiffusion synthetic data to real data | Mask2Former ResNet50 · MS-COCO | mIoU: 57.8 -> 58.5 | Improved |
| add JoDiffusion synthetic data to real data | DeepLabV3 ResNet50 · MS-COCO | mIoU: 48.9 -> 56.4 | Improved |
| add SDS synthetic data to real data | DeepLabV3 ResNet101 · Pascal VOC | mIoU: 79.9 -> 79.8 | No clear effect |
| use JoDiffusion instead of FreeMask | Mask2Former Swin-S · ADE20K | mIoU: 52.1 -> 52.2 | No clear effect |
| use JoDiffusion instead of FreeMask | Mask2Former ResNet50 · Pascal VOC | mIoU: 77.9 -> 79.4 | Improved |
Earlier methods this paper compares with
Named baselines from earlier accepted papers in this paper's results tables: better than 1 of 1 on most shared results.
- Dataset Diffusion — Dataset Diffusion: Diffusion-based Synthetic Data Generation for Pixel-Level Semantic Segmentation (NeurIPS 2023) · better on 10 of 10 shared results
Matched by method name in results tables read by AI. Authors' own tables tend to favour the newer method, and a comparison can be missed when a method is renamed.
Closest papers by meaning
- EmerDiff: Emerging Pixel-level Semantic Knowledge in Diffusion Models (ICLR 2024 · similarity 0.91)
- Seg4Diff: Unveiling Open-Vocabulary Semantic Segmentation in Text-to-Image Diffusion Transformers (NeurIPS 2025 · similarity 0.90)
- Unleashing the Potential of the Diffusion Model in Few-shot Semantic Segmentation (NeurIPS 2024 · similarity 0.88)
- Open-Vocabulary Attention Maps with Token Optimization for Semantic Segmentation in Diffusion Models (CVPR 2024 · similarity 0.88)
- VidSeg: Training-free Video Semantic Segmentation based on Diffusion Models (CVPR 2025 · similarity 0.87)
- High-Precision Dichotomous Image Segmentation via Probing Diffusion Capacity (ICLR 2025 · 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.
