Research map / Detection and segmentation / CNN features
Zone Evaluation: Revealing Spatial Bias in Object Detection
Zhaohui Zheng, Yuming Chen, Qibin Hou, Xiang Li, Ping Wang, Ming‐Ming Cheng
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 SELA (spatial equilibrium label assignment) | GFocal R-18 · VOC 07+12 | ZP variance: 53.6 -> 37.7 | Improved |
| add SELA | GFocal R-50 · COCO 2017 | ZP variance: 16.9 -> 14.4 | Improved |
| add SELA | GFocal R-18 · Fruit | ZP variance: 56.2 -> 33.6 | Improved |
| add SELA | GFocal R-18 · Helmet | ZP variance: 3.0 -> 3.1 | No clear effect |
| increase gamma in SELA from 0.2 to 0.4 (ablation) | GFocal R-18 · VOC 07+12 | AP: 52.8 -> 52.0 | Hurt |
| raise gamma in SELA from 0.1 to 0.2 (ablation) | GFocal R-18 · VOC 07+12 | ZP variance: 44.5 -> 37.7 | Improved |
| use position-dependent spatial weight instead of constant 1 (ablation) | SELA on GFocal R-18 · VOC 07+12 | ZP variance: 48.3 -> 37.7 | Improved |
| add SELA with ResNet-50 backbone | GFocal R-50 · VOC 07+12 | ZP variance: 41.5 -> 32.2 | Improved |
| add SE loss | GFocal R-18 · VOC 07+12 | ZP variance: 53.6 -> 41.6 | Improved |
| add SE loss to DINO | DINO · VOC 07+12 | ZP variance: 47.6 -> 46.7 | Improved |
| use full-conv padding in head (ablation) | GFocal R-50 · COCO val2017 | AP and ZP variance: AP 40.1 -> 38.6; variance 16.9 -> 12.4 | Mixed |
| remove oversized anchors outside the image (ablation) | GFocal R-50 · COCO val2017 | ZP variance: 16.9 -> 16.7 | No clear effect |
Closest papers by meaning
- Exploring the Bounds of the Utility of Context for Object Detection (CVPR 2019 · similarity 0.87)
- Localization Recall Precision (LRP): A New Performance Metric for Object Detection (ECCV 2018 · similarity 0.86)
- Pixel-level Quality Assessment for Oriented Object Detection (AAAI 2026 · similarity 0.86)
- DOTA: A Large-Scale Dataset for Object Detection in Aerial Images (CVPR 2018 · similarity 0.86)
- Analysing Domain Shift Factors between Videos and Images for Object Detection (TPAMI 2016 · similarity 0.86)
- Benchmarking Object Detectors under Real-World Distribution Shifts in Satellite Imagery (CVPR 2025 · 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.
