Research map / Detection and segmentation / Open-set and few-shot detection
Interactive Image Segmentation With Latent Diversity
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 |
|---|---|---|---|
| add the latent-diversity proposal and selection architecture | interactive image segmentation · DAVIS, GrabCut, SBD, and COCO | clicks to reach 85% or 90% IoU | Helped |
| use VGG feature maps in the augmented input (ablation) | the segmentation network · DAVIS | average clicks to 85% IoU: 9.41 -> 5.95 | Helped |
| use IoU loss instead of cross-entropy (ablation) | the segmentation network · DAVIS | average clicks to 85% IoU: 7.18 -> 5.95 | Helped |
| use IoU loss instead of L1 (ablation) | the segmentation network · DAVIS | average clicks to 85% IoU: 6.93 -> 5.95 | Helped |
| generate diverse intermediate segmentations (ablation) | the segmentation network · DAVIS | average clicks to 85% IoU: 7.26 -> 5.95 | Helped |
| include distance transforms in the input (ablation) | the segmentation network · DAVIS | average clicks to 85% IoU: 9.11 -> 5.95 | Helped |
| include positive and negative clicks in the input (ablation) | the segmentation network · DAVIS | average clicks to 85% IoU: 9.67 -> 5.95 | Helped |
| add click soft constraints to the loss (ablation) | the segmentation network · DAVIS | average clicks to 85% IoU: 6.32 -> 5.95 | Helped |
| use a selection network rather than ranked diversity loss (ablation) | the proposal-selection pipeline · DAVIS | average clicks to 85% IoU: 6.13 -> 5.95 | Helped |
| use CAN rather than U-net for the segmentation network (ablation) | the proposed approach · DAVIS and GrabCut | clicks to reach 85% or 90% IoU: 6.25 -> 5.95; 9.96 -> 9.57; 3.22 -> 3.20; 4.81 -> 4.79 | Helped |
| increase the number of intermediate solutions up to M = 6 (ablation) | the proposal network · DAVIS and GrabCut | clicks to reach 85% or 90% IoU | Helped |
| remove graph-cut postprocessing | the DOS baseline · VGA-resolution images | running time (ms): 549 -> 236 | Helped |
Closest papers by meaning
- Rethinking Interactive Image Segmentation with Low Latency High Quality and Diverse Prompts (CVPR 2024 · similarity 0.90)
- TETRIS: Towards Exploring the Robustness of Interactive Segmentation (AAAI 2024 · similarity 0.89)
- NTClick: Achieving Precise Interactive Segmentation With Noise-tolerant Clicks (CVPR 2025 · similarity 0.86)
- Explore In-Context Segmentation via Latent Diffusion Models (AAAI 2025 · similarity 0.86)
- RClicks: Realistic Click Simulation for Benchmarking Interactive Segmentation (NeurIPS 2024 · similarity 0.86)
- Repurposing Stable Diffusion Attention for Training-Free Unsupervised Interactive Segmentation (CVPR 2025 · similarity 0.86)
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.