Research map / Detection and segmentation / CNN features
Split-GCN: Effective Interactive Annotation for Segmentation of Disconnected Instance
Namgil Kim, Barom Kang, Yeonok Cho
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 motion vector branch to GCN (ablation) | GCN baseline · Cityscapes val | mIoU: 70.8 -> 73.1 | Improved |
| add separating network after motion branch (ablation) | GCN + motion branch · Cityscapes val | mIoU: 73.1 -> 76.6 | Improved |
| add interactive learning (ablation) | Split-GCN · Cityscapes val | mIoU: 76.6 -> 78.3 | Improved |
| increase split factor from 1 to 3 (ablation) | Split-GCN · Cityscapes val | mIoU: 75.3 -> 76.6 | Improved |
| increase split factor beyond 3 (k=5) (ablation) | Split-GCN · Cityscapes val | mIoU: 76.6 -> 75.9 | Hurt |
| increase control points from 20 to 40 (ablation) | Split-GCN · Cityscapes val | mIoU: 71.3 -> 76.6 | Improved |
| increase control points from 40 to 60 (ablation) | Split-GCN · Cityscapes val | mIoU: 76.6 -> 76.2 | No clear effect |
| use separating network with k=3 vs k=1 on cross-domain Rooftop/Card.MR | Split-GCN · Rooftop, Card.MR | mIoU: 80.1 -> 80.2; 92.0 -> 92.3 | Improved |
| Split-GCN vs Spline-GCN in automatic mode | Split-GCN (k=3) · Cityscapes val | mIoU: 73.7 -> 76.6 | Improved |
| add one click in interactive mode | Split-GCN · Cityscapes val | mIoU: 78.3 -> 79.4 | Improved |
| add two clicks in interactive mode | Split-GCN · Cityscapes val | mIoU: 78.3 -> 81.2 | Improved |
Earlier methods this paper compares with
Named baselines from earlier accepted papers in this paper's results tables: better than 0 of 2 on most shared results.
- SSAP — SSAP: Single-Shot Instance Segmentation With Affinity Pyramid (ICCV 2019) · better on 0 of 1 shared result
- AdaptIS — AdaptIS: Adaptive Instance Selection Network (ICCV 2019) · better on 0 of 1 shared result
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
- Fast Interactive Object Annotation With Curve-GCN (CVPR 2019 · similarity 0.88)
- Annotating Object Instances With a Polygon-RNN (CVPR 2017 · similarity 0.88)
- Efficient Interactive Annotation of Segmentation Datasets With Polygon-RNN++ (CVPR 2018 · similarity 0.87)
- SUM Parts: Benchmarking Part-Level Semantic Segmentation of Urban Meshes (CVPR 2025 · similarity 0.82)
- The Devil is in the Object Boundary: Towards Annotation-free Instance Segmentation using Foundation Models (ICLR 2024 · similarity 0.82)
- PartNeXt: A Next-Generation Dataset for Fine-Grained and Hierarchical 3D Part Understanding (NeurIPS 2025 · similarity 0.82)
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.
