Oriented R-CNN for Object Detection

By Xingxing Xie, Gong Cheng, Jiabao Wang, Xiwen Yao, Junwei Han. IEEE International Conference on Computer Vision, 2021

1,161 citations recorded by the source corpus.

Current state-of-the-art two-stage detectors generate oriented proposals through time-consuming schemes. This diminishes the detectors’ speed, thereby becoming the computational bottleneck in advanced oriented object detection systems. This work proposes an effective and simple oriented object detection framework, termed Oriented R-CNN, which is a general two-stage oriented detector with promising accuracy and efficiency. To be specific, in the first stage, we propose an oriented Region Proposal Network (oriented RPN) that directly generates high-quality oriented proposals in a nearly cost-free manner. The second stage is oriented R-CNN head for refining oriented Regions of Interest (oriented RoIs) and recognizing them. Without tricks, oriented R-CNN with ResNet50 achieves state-of-the-art detection accuracy on two commonly-used datasets for oriented object detection including DOTA (75.87% mAP) and HRSC2016 (96.50% mAP), while having a speed of 15.1 FPS with the image size of 1024×1024 on a single RTX 2080Ti. We hope our work could inspire rethinking the design of oriented detectors and serve as a baseline for oriented object detection. Code is available at https://github.com/jbwang1997/OBBDetection.

Subfields: Artificial Intelligence, Computer Vision and Pattern Recognition, Neurology

View Paper: https://www.semanticscholar.org/paper/f0035ac7656226b653d475e57a1397c4ce7106b4

How researchers see this paper

The researchers whose trusted networks rank it highest, strongest advocate first. Every researcher gets a different order.

In Artificial Intelligence

1. Dacheng Tao
2. Jing Zhang
3. Muhammad Hussain
4. Bo Du
5. Qiming Zhang
6. Yufei Xu
7. Di Wang
8. Guisong Xia

In Computer Vision and Pattern Recognition

1. Junwei Han
2. Gong Cheng
3. Dacheng Tao
4. Jianke Zhu
5. Wentong Li
6. Guisong Xia
7. Jing Zhang
8. Bo Du

In Neurology

1. Muhammad Hussain
2. Dacheng Tao
3. Heng Fan
4. Xue Yang
5. Bo Du
6. Ke Yan
7. Kai Chen
8. Jianke Zhu

Citations in the corpus

• RingMo: A Remote Sensing Foundation Model With Masked Image Modeling (2023, neutral): “Restrictions apply. dataset, named OBB [98], [100].”
• YOLOv1 to v8: Unveiling Each Variant–A Comprehensive Review of YOLO (2024, methodology): “Examples of two-stage detectors include RCNN (Region-based Convolutional Neural Network) [19], Fast R-CNN [20], Faster R-CNN [21], and FPN (Feature Pyramid Network) [22].”
• Poly Kernel Inception Network for Remote Sensing Detection (2024, methodology): “Top : Our approach yields solid performance gains over various remote sensing detectors [10, 20, 59, 65, 71] with fewer parameters on DOTA-v1.”
• Advancing Plain Vision Transformer Toward Remote Sensing Foundation Model (2022, methodology): “As can be seen, based on the ViTAE backbone equipped with the proposed RVSA modules, Oriented R-CNN can accurately recognize and locate various artificial or natural objects with high confidence no matter in city and rural scenes or dense and sparse distributions.”
• An Empirical Study of Remote Sensing Pretraining (2022, methodology): “We adopt the default hyper-parameters of ORCN, which is implemented in OBBDetection2.”
• ICAFusion: Iterative Cross-Attention Guided Feature Fusion for Multispectral Object Detection (2023, background): “As a fundamental vision task, object detection remains a hot topic in both academia and industry, and has made considerable progress in small object detection [1], face and pedestrian detection [2] and oriented object detection [3, 4], thanks to the rapid development of convolutional neural networks”
• Oriented RepPoints for Aerial Object Detection (2021, methodology): “SCRDet [45], CAD-Net [47], DRN [24], R3Det [42], ReDet [9] and Oriented R-CNN [36] achieve the promising performance by predicting the rotation angles of bounding boxes.”
• Anchor-Free Oriented Proposal Generator for Object Detection (2021, background): “Benefiting from the development of generic detection, oriented object detection also has made great progress [1]–[3], [31]–[39] in recent years.”
• MMRotate: A Rotated Object Detection Benchmark using PyTorch (2022, background): “OBBDetection 4 [30] is another popular open source oriented object detection toolbox.”
• Towards Large-Scale Small Object Detection: Survey and Benchmarks (2022, background): “Oriented RCNN [177] obtains a relatively high performance both at overall AP (34.4%).”