RECORD
D2Q-DETR: Decoupling and Dynamic Queries for Oriented Object Detection with Transformers
Computer Vision and Pattern Recognition · Submitted 1 Mar 2023
abstract compiled · 2026-09-23 · no language-model review
The authors' main claim Experiments on the largest and challenging DOTA-v1.0 and DOTA-v1.5 datasets show that D2Q-DETR outperforms existing NMS-based and NMS-free oriented object detection methods and achieves the new state-of-the-art.
The compiled record
The paper's own sentences, arranged. Each label is drawn from a closed vocabulary; the number beside it is the probability the classifier gave that label. Numbers in the tables are the authors', checked by code. No sentence in this record was written by a model.
Open the compiled record
The abstract, sentence by sentence
Despite the promising results, existing oriented object detection methods usually involve heuristically designed rules, e.g., RRoI generation, rotated NMS. In this paper, we propose an end-to-end framework for oriented object detection, which simplifies the model pipeline and obtains superior performance. Our framework is based on DETR, with the box regression head replaced with a points prediction head. The learning of points is more flexible, and the distribution of points can reflect the angle and size of the target rotated box. We further propose to decouple the query features into classification and regression features, which significantly improves the model precision. Aerial images usually contain thousands of instances. To better balance model precision and efficiency, we propose a novel dynamic query design, which reduces the number of object queries in stacked decoder layers without sacrificing model performance. Finally, we rethink the label assignment strategy of existing DETR-like detectors and propose an effective label re-assignment strategy for improved performance. We name our method D2Q-DETR. Experiments on the largest and challenging DOTA-v1.0 and DOTA-v1.5 datasets show that D2Q-DETR outperforms existing NMS-based and NMS-free oriented object detection methods and achieves the new state-of-the-art.
background contribution method result interpretation main claim
| # | sentence | role | the authors' own work | states the main conclusion | strength |
|---|---|---|---|---|---|
| s0 | Despite the promising results, existing oriented object detection methods usually involve heuristically designed rules, e.g., RRoI generation, rotated NMS. | background 0.66 | 0.05 | 0.02 | asserted 0.97 |
| s1 | In this paper, we propose an end-to-end framework for oriented object detection, which simplifies the model pipeline and obtains superior performance. | contribution 1.00 | 0.90 | 0.29 | asserted 1.00 |
| s2 | Our framework is based on DETR, with the box regression head replaced with a points prediction head. | method 0.94 | 0.91 | 0.04 | asserted 1.00 |
| s3 | The learning of points is more flexible, and the distribution of points can reflect the angle and size of the target rotated box. | method 0.74 | 0.50 | 0.04 | asserted 0.99 |
| s4 | We further propose to decouple the query features into classification and regression features, which significantly improves the model precision. | contribution 0.75 | 0.94 | 0.07 | asserted 1.00 |
| s5 | Aerial images usually contain thousands of instances. | background 1.00 | 0.11 | 0.01 | asserted 1.00 |
| s6 | To better balance model precision and efficiency, we propose a novel dynamic query design, which reduces the number of object queries in stacked decoder layers without sacrificing model performance. | contribution 0.87 | 0.94 | 0.07 | asserted 0.99 |
| s7 | Finally, we rethink the label assignment strategy of existing DETR-like detectors and propose an effective label re-assignment strategy for improved performance. | contribution 0.91 | 0.93 | 0.04 | asserted 0.98 |
| s8 | We name our method D2Q-DETR. | contribution 0.80 | 0.79 | 0.02 | asserted 0.88 |
| s9 | Experiments on the largest and challenging DOTA-v1.0 and DOTA-v1.5 datasets show that D2Q-DETR outperforms existing NMS-based and NMS-free oriented object detection methods and achieves the new state-of-the-art. | result 1.00 | 0.96 | 0.88 | asserted 1.00 |
What was examined, and what was not
- arXiv listing abstract v1
- full text
- figures and tables
- references and citing sentences
- relations to other papers
- P abstract
- S sections
- T body sentences
- C citations
- R cross-paper relations
The paper
Original abstract
Despite the promising results, existing oriented object detection methods usually involve heuristically designed rules, e.g., RRoI generation, rotated NMS. In this paper, we propose an end-to-end framework for oriented object detection, which simplifies the model pipeline and obtains superior performance. Our framework is based on DETR, with the box regression head replaced with a points prediction head. The learning of points is more flexible, and the distribution of points can reflect the angle and size of the target rotated box. We further propose to decouple the query features into classification and regression features, which significantly improves the model precision. Aerial images usually contain thousands of instances. To better balance model precision and efficiency, we propose a novel dynamic query design, which reduces the number of object queries in stacked decoder layers without sacrificing model performance. Finally, we rethink the label assignment strategy of existing DETR-like detectors and propose an effective label re-assignment strategy for improved performance. We name our method D2Q-DETR. Experiments on the largest and challenging DOTA-v1.0 and DOTA-v1.5 datasets show that D2Q-DETR outperforms existing NMS-based and NMS-free oriented object detection methods and achieves the new state-of-the-art.
Human review
Discussion
Know something about this paper? Leave a comment, or highlight any passage above to comment on it.