Remote Sensing Super-resolution Object Detection (RSSOD) Dataset
Description
A new benchmark dataset RSSOD, for object detection of small objects in remote sensing images, with a high overlap of classes. Total 3 sets of annotations, 5-class, 4, class, 2-class, and Single-Class. Format: YOLO and COCO. If this dataset helped you in your research please cite: 1- Bashir, Syed Muhammad Arsalan; Wang, Yi; Khan, Mahrukh; Ullah, Qudrat; Wang, Rui; Song, Yilin; Guo, Zhe; Niu, Yilong (2021), “Remote Sensing Super-resolution Object Detection (RSSOD)”, Mendeley Data, V1, DOI: 10.17632/b268jv86tf.1 2- Wang, Yi; Bashir, Syed Muhammad Arsalan; Khan, Mahrukh; Ullah, Qudrat; Wang, Rui; Song, Yilin; Guo, Zhe; Niu, Yilong (2021), “Remote Sensing Image Super-resolution and Object Detection: Benchmark and State of the Art ”, arXiv. https://arxiv.org/abs/2111.03260
Files
Steps to reproduce
Use MATLAB "imresize ( )" function with bicubic interpolation to generate LR images with different scale factors. The annotations files are in YOLO and COCO format.