A Small Object Detection Solution by Using Super- Resolution Recovery
Xing Chen, Liang Xi, Zhiyan Bao · 2019
Small object detection is the key challenge to aerial image object detection. Small objects in aerial images are usually too fuzzy to detect. This paper proposes a solution by using Generative Adversarial Network to recover small objects' low-resolution image to high-resolution to get better detection performance. Experiment result shows this solution achieves mAP 68.38% and missing rate 14.23%.