Our new X account is live! Follow @wizwand_team for updates
WorkDL logo mark

Oriented R-CNN for Object Detection

About

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$\times$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.

Xingxing Xie, Gong Cheng, Jiabao Wang, Xiwen Yao, Junwei Han• 2021

Related benchmarks

TaskDatasetResultRank
Oriented Object DetectionDOTA v1.0 (test)
SV82.2
378
Object DetectionDOTA 1.0 (test)
Plane AP90.26
256
Oriented Object DetectionDOTA (test)--
92
Object DetectionHRSC 2016 (test)
mAP@0.0790.5
72
Object DetectionDroneVehicle (test)
mAP5067
61
Oriented Object DetectionHRSC 2016 (test)--
55
Object DetectionDIOR
mAP5064.38
50
Rotated Object DetectionDOTA 1.0 (test)
mAP80.87
46
Oriented Object DetectionDIOR-R
mAP5064.3
44
Object DetectionDOTA v1.5
mAP39.31
37
Showing 10 of 31 rows

Other info

Code

Follow for update