Geospatial Object Detection on High Resolution Remote Sensing Imagery Based on Double Multi-Scale Feature Pyramid Network

[업데이트 2019.08.24 14:15]

 

1. 논문

Geospatial Object Detection on High Resolution Remote Sensing Imagery Based on Double Multi-Scale Feature Pyramid Network 

 

Remote Sens. 2019, 11(7), 755; https://doi.org/10.3390/rs11070755

Received: 17 February 2019 / Revised: 20 March 2019 / Accepted: 23 March 2019 / Published: 28 March 2019

 

2. 요약

- To solve these problems, an effective region-based VHR remote sensing imagery object detection framework named Double Multi-scale Feature Pyramid Network (DM-FPN) was proposed in this paper, which utilizes inherent multi-scale pyramidal features and combines the strong-semantic, low-resolution features and the weak-semantic, high-resolution features simultaneously.