Robust Data Augmentation Generative Adversarial Network for Object Detection

Published in Sensors, 2023

We propose a GAN architecture that generates training images together with their bounding-box labels, so that scarce object-detection datasets (e.g., fire detection) can be augmented with fully labeled synthetic data.

Recommended citation: Hyungtak Lee, Seongju Kang, and Kwangsue Chung. (2023). "Robust Data Augmentation Generative Adversarial Network for Object Detection." Sensors, 23(1), 157. https://doi.org/10.3390/s23010157
Download Paper