Research map / Detection and segmentation / Open-set and few-shot detection
COCO-Stuff: Thing and Stuff Classes in Context
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 |
|---|---|---|---|
| Increase DeepLab V2 training data from 1K to 118K images | DeepLab V2 with VGG-16 · COCO-Stuff val | mean IOU: 15.9% -> 33.2% | Helped |
| Evaluate thing classes instead of stuff classes | DeepLab V2 with VGG-16 · COCO-Stuff val | mean IOU: 24.0% -> 43.6% | Helped |
| Use superpixel annotation instead of freedraw annotation | COCO-Stuff annotation protocol · COCO-Stuff annotation images | speedup: 2.8 | Helped |
| Use superpixel annotation instead of freedraw annotation | COCO-Stuff annotation protocol · COCO-Stuff annotation images | reference agreement: 96.1% | No clear effect |
| Add COCO thing overlays to the annotation process (ablation) | Superpixel annotation · COCO annotation images | speedup: 1.2x | Helped |
| Add COCO thing overlays to the annotation process (ablation) | Superpixel annotation · COCO annotation images | reference agreement: 96.1% in both cases | No clear effect |
| Use freedraw instead of superpixel annotation | COCO-Stuff annotation protocol · COCO annotation images | speedup: 2.8 | Hurt |
Closest papers by meaning
- COCONut: Modernizing COCO Segmentation (CVPR 2024 · similarity 0.86)
- The Devil is in the Fine-Grained Details: Evaluating Open-Vocabulary Object Detectors for Fine-Grained Understanding (CVPR 2024 · similarity 0.86)
- The Devil is in the Object Boundary: Towards Annotation-free Instance Segmentation using Foundation Models (ICLR 2024 · similarity 0.84)
- Compositional Caching for Training-free Open-vocabulary Attribute Detection (CVPR 2025 · similarity 0.84)
- COCOTree: A Dataset and Benchmark for Open Tree-Structured Visual Decomposition (NeurIPS 2026 · similarity 0.84)
- Common Inpainted Objects In-N-Out of Context (CVPR 2026 · similarity 0.84)
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.