English | 简体中文
A curated collection of oriented object detection papers, code, and reading notes, covering detection frameworks, angle coding, Gaussian representations, rotation-aware features, real-time DETR, point supervision, and language guidance.
English is the default language. A Simplified Chinese version is available above.
Last verified: 2026-09-12 · 15 papers · 3 abstract screening notes
Reading tracker · Note template · Contributing · Changelog
- DETR and Real-Time Detection
- Gaussian Representations and Regression Losses
- Angle Coding
- Rotation-Aware Features and Multi-Scale Representations
- Oriented Detection Frameworks
- Point Supervision and Annotation Efficiency
- Language Guidance
- Datasets and Tools
- About This Collection
- References and Acknowledgments
Papers are sorted by year within each topic, newest first, with one primary entry per paper. paper links to the original source, code to the authors’ implementation, and note to our English reading note. Notes distinguish abstract screening, close reading, and reproduction; the current notes are at the abstract screening stage.
| Year | Pub | Title | Links |
|---|---|---|---|
| 2026 | ECCV accepted¹ | RiO-DETR: DETR for Real-time Oriented Object Detection | paper · code: partial |
| 2026 | IEEE TGRS | Real-Time Oriented Object Detection Transformer in Remote Sensing Images | paper · doi · code · note |
| Year | Pub | Title | Links |
|---|---|---|---|
| 2025 | CVPR | GauCho: Gaussian Distributions with Cholesky Decomposition for Oriented Object Detection | paper · code · note |
| 2021 | NeurIPS | Learning High-Precision Bounding Box for Rotated Object Detection via Kullback-Leibler Divergence | paper · code |
| 2021 | ICML | Rethinking Rotated Object Detection with Gaussian Wasserstein Distance Loss | paper · code · note |
| Year | Pub | Title | Links |
|---|---|---|---|
| 2023 | CVPR | Phase-Shifting Coder: Predicting Accurate Orientation in Oriented Object Detection | paper · code |
| 2020 | ECCV | Arbitrary-Oriented Object Detection with Circular Smooth Label | paper · code |
| Year | Pub | Title | Links |
|---|---|---|---|
| 2026 | arXiv preprint | FGAA-FPN: Foreground-Guided Angle-Aware Feature Pyramid Network for Oriented Object Detection | paper · code: unverified |
| 2023 | ICCV | Adaptive Rotated Convolution for Rotated Object Detection | paper · code |
| 2021 | CVPR | ReDet: A Rotation-Equivariant Detector for Aerial Object Detection | paper · code |
| Year | Pub | Title | Links |
|---|---|---|---|
| 2021 | ICCV | Oriented R-CNN for Object Detection | paper · code |
| 2019 | CVPR | Learning RoI Transformer for Oriented Object Detection in Aerial Images | paper · code |
| Year | Pub | Title | Links |
|---|---|---|---|
| 2026 | ICLR | Point2RBox-v3: Self-Bootstrapping from Point Annotations via Integrated Pseudo-Label Refinement and Utilization | paper · proceedings · code |
| 2025 | CVPR | Point2RBox-v2: Rethinking Point-supervised Oriented Object Detection with Spatial Layout Among Instances | paper · code |
| Year | Pub | Title | Links |
|---|---|---|---|
| 2025 | SOICT accepted² | LOGOS: Language-guided Oriented Object Detection in Aerial Scenes | paper · code: unverified |
¹ RiO-DETR: ECCV 2026 acceptance is stated in the authors’ arXiv comments. The repository currently provides the base RT-DETR-OBB framework and some weights, without the paper-specific technical improvements. The complete research code has not been released. Authors’ notice
² LOGOS: First posted to arXiv on 2026-07-09, with author comments stating acceptance at SOICT 2025. It is listed under the stated acceptance year; the conference proceedings have not been independently checked. Source metadata
Point2RBox-v3 first appeared in 2025 and is listed under ICLR 2026. FGAA-FPN remains a preprint: a submission statement is not evidence of acceptance. Other code links identify author-provided implementations; they do not indicate successful reproduction here.
| Resource | Purpose |
|---|---|
| DOTA | Public datasets, OBB annotation documentation, and evaluation resources |
| MMRotate | Oriented detection toolbox, implementations, and experiment configurations |
| RotationDetection | Author implementations of GWD, KLD, and related methods |
- This is a curated starter collection maintained manually in Markdown. We check new papers alongside acceptance and code updates for existing entries. Automated collection is not configured.
- Publication metadata comes primarily from conference websites, proceedings, arXiv, and author repositories. Preprints and acceptance claims supported only by author statements are labeled explicitly.
- See TRACKING.md for reading and reproduction status. Experiment records include data splits, backbone, tiling, input size, hardware, precision, and timing boundaries.
- See CONTRIBUTING.md to add papers or notes. A suggested reading order is Oriented R-CNN for the framework, then GWD / KLD / CSL / PSC, followed by a comparison of O2-RT-DETR and RiO-DETR.
The topic tables and standalone Markdown notes are inspired by whut09/Awesome-object-detection. Our screening notes are written from the original paper sources. See SOURCES.md for discovery resources and other layout references.
Original content in this repository is available under the MIT License. Linked papers, external code, and datasets retain their respective licenses.