Research map / Image generation / Text-to-image diffusion
LD-ZNet: A Latent Diffusion Approach for Text-Based Image Segmentation
Koutilya PNVR, Bharat Singh, Pallabi Ghosh, Behjat Siddiquie, David Jacobs
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 |
|---|---|---|---|
| use LDM latent z as input instead of RGB image (ZNet vs RGBNet) (ablation) | text-based segmentation network · PhraseCut test | mIoU: 46.7 -> 51.3 | Improved |
| add LDM internal diffusion features via cross-attention (LD-ZNet vs ZNet) (ablation) | ZNet · PhraseCut test | mIoU: 51.3 -> 52.7 | Improved |
| use z-space input instead of RGB (ablation) | ZNet · AIGI | mIoU: 63.4 -> 68.4 | Improved |
| add LDM diffusion features on top of ZNet (ablation) | ZNet · AIGI | mIoU: 68.4 -> 74.1 | Improved |
| inject LDM features via cross-attention instead of concatenation (ablation) | LD-ZNet · PhraseCut test | mIoU: 50.2 -> 52.7 | Improved |
| use z input and LDM features (ablation) | LD-ZNet · RefCOCO (zero-shot from PhraseCut) | IoU: 36.3 -> 41.0 | Improved |
| use LDM-initialized UNet segmentation network (RGBNet) vs CLIPSeg/MDETR/SEEM | RGBNet · AIGI | mIoU: 63.4 vs 57.4 (SEEM) | Improved |
| use LDM features from middle blocks 6-10 at timesteps 300-500 (ablation) | LD-ZNet feature selection · PhraseCut small validation subset | AP | Improved |
Closest papers by meaning
- Unleashing the Potential of the Diffusion Model in Few-shot Semantic Segmentation (NeurIPS 2024 · similarity 0.89)
- Explore In-Context Segmentation via Latent Diffusion Models (AAAI 2025 · similarity 0.89)
- Seg4Diff: Unveiling Open-Vocabulary Semantic Segmentation in Text-to-Image Diffusion Transformers (NeurIPS 2025 · similarity 0.88)
- Text-Image Alignment for Diffusion-Based Perception (CVPR 2024 · similarity 0.88)
- EmerDiff: Emerging Pixel-level Semantic Knowledge in Diffusion Models (ICLR 2024 · similarity 0.88)
- VidSeg: Training-free Video Semantic Segmentation based on Diffusion Models (CVPR 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.
