Research map / Detection and segmentation / CNN features
Boundary IoU: Improving Object-Centric Image Segmentation Evaluation
Check 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 |
|---|---|---|---|
| Use Boundary AP instead of Mask AP to evaluate low-resolution synthetic masks | Synthetic 28×28 predictions on COCO val · COCO val synthetic 28×28 | AP_L: 95.0 -> 73.0 | Helped |
| Use Boundary PQ instead of Mask PQ for low-resolution predictions | Synthetic panoptic predictions on Cityscapes · Cityscapes val synthetic downscaling ratio 8 | SQ: 77.8 -> 58.6 | Helped |
| Use Boundary PQ instead of Mask PQ | Synthetic panoptic predictions on Cityscapes · Cityscapes val synthetic downscaling ratio 8 | RQ: 83.7 -> 80.2 | No clear effect |
| Increase PointRend output resolution from 28×28 to 224×224 | PointRend on COCO val · COCO val | Boundary AP: 41.6 -> 42.1 | Helped |
| Add boundary supervision and higher-resolution mask-head features | Mask R-CNN on COCO val · COCO val | Boundary AP: 36.1 -> 42.3 | Helped |
| Use 224×224 PointRend outputs rather than BMask R-CNN's 28×28 outputs | PointRend and BMask R-CNN on COCO val · COCO val large objects | Boundary AP_L: 33.9 -> 34.4 | Helped |
| Use BMask R-CNN rather than PointRend | Instance segmentation on COCO val · COCO val small objects | Boundary AP_S: 48.3 -> 48.7 | Helped |
| Replace Mask IoU with Boundary IoU in the evaluation measure | Large-object segmentation quality evaluation · COCO val synthetic 28×28 | AP_L: 95.0 -> 73.0 | Helped |
Closest papers by meaning
- Revisiting Evaluation Metrics for Semantic Segmentation: Optimization and Evaluation of Fine-grained Intersection over Union (NeurIPS 2023 · similarity 0.87)
- Generalized Intersection Over Union: A Metric and a Loss for Bounding Box Regression (CVPR 2019 · similarity 0.87)
- Rethinking Boundary Discontinuity Problem for Oriented Object Detection (CVPR 2024 · similarity 0.85)
- Uncertainty-aware Fine-tuning of Segmentation Foundation Models (NeurIPS 2024 · similarity 0.84)
- RankSEG-RMA: An Efficient Segmentation Algorithm via Reciprocal Moment Approximation (NeurIPS 2025 · similarity 0.84)
- unMORE: Unsupervised Multi-Object Segmentation via Center-Boundary Reasoning (ICML 2025 · similarity 0.83)
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.