Research map / Detection and segmentation / Weakly supervised detection and segmentation
H2RBox: Horizontal Box Annotation is All You Need for Oriented Object Detection
Xue Yang, Gefan Zhang, Wentong Li, Yue Zhou, Xuehui Wang, Junchi Yan
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 |
|---|---|---|---|
| Train H2RBox from horizontal boxes using two-view self-supervised angle consistency | FCOS-based oriented detector · DOTA-v1.0 | AP50: 67.90 vs. 53.59 | Improved |
| Train H2RBox from horizontal boxes using two-view self-supervised angle consistency | FCOS-based oriented detector · DIOR-R test | AP75: 32.60 vs. 28.10 | Improved |
| Use H2RBox instead of RBox-supervised FCOS | FCOS-based oriented detector · DOTA-v1.0, multi-scale | AP50: 75.31 vs. 74.40 | Hurt |
| Use H2RBox instead of RBox-supervised FCOS | FCOS-based oriented detector · DIOR-R test | AP: 33.15 vs. 34.16 | Hurt |
| Use reflection padding rather than zeros padding for view generation (ablation) | H2RBox · DOTA-v1.0 | AP: 20.17 -> 35.92 | Improved |
| Add center-region cropping to reflection padding (ablation) | H2RBox · DOTA-v1.0 | AP: 35.92 -> 33.60 | Hurt |
| Use one-to-one rather than one-to-many label reassignment (ablation) | H2RBox · DOTA-v1.0 | AP: 21.60 -> 35.92 | Improved |
| Use one-to-one rather than one-to-many label reassignment (ablation) | H2RBox · DIOR-R | AP: 31.10 -> 33.15 | Improved |
| Add self-supervised consistency loss (ablation) | H2RBox · DOTA-v1.0 | AP: 12.63 -> 35.92 | Improved |
| Add self-supervised consistency loss (ablation) | H2RBox · DIOR-R | AP: 15.27 -> 33.15 | Improved |
| Mask the SS loss for circular classes during training (ablation) | H2RBox · DOTA-v1.0 | AP: 31.90 -> 36.36 | Improved |
| Use horizontal circumscribed rectangles as circular-class outputs at test time (ablation) | H2RBox · DOTA-v1.0 | AP: 31.90 -> 35.92 | Improved |
Closest papers by meaning
- ABBSPO: Adaptive Bounding Box Scaling and Symmetric Prior based Orientation Prediction for Detecting Aerial Image Objects (CVPR 2025 · similarity 0.89)
- Point2RBox: Combine Knowledge from Synthetic Visual Patterns for End-to-end Oriented Object Detection with Single Point Supervision (CVPR 2024 · similarity 0.89)
- PointOBB-v2: Towards Simpler, Faster, and Stronger Single Point Supervised Oriented Object Detection (ICLR 2025 · similarity 0.87)
- BGHR: Bridging the Gap Between HBox-Supervised and RBox-Supervised Oriented Object Detection via Adaptive Fine-Grained Sample Mining (AAAI 2025 · similarity 0.87)
- WeakPCSOD: Overcoming the Bias of Box Annotations for Weakly Supervised Point Cloud Salient Object Detection (AAAI 2024 · similarity 0.86)
- Point2RBox-v2: Rethinking Point-supervised Oriented Object Detection with Spatial Layout Among Instances (CVPR 2025 · similarity 0.85)
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.
