InfraRed Images Augmentation Based on Images Generation with Generative Adversarial Networks
Foji Chen, Feng Zhu, Qingxiao Wu, Yingming Hao, Yunge Cui, Ende Wang · 2019 IEEE International Conference on Unmanned Systems (ICUS) · 2019
The challenge brought by small data sets is a common problem in many computer visual tasks. The method of images generation with generative adversarial networks is investigated as a solution to extend small infrared datasets in this paper. In such a solution, the generation of infrared images is realized by generative adversarial networks by which RGB images can be translated to infrared images. The images generated by the generator of generative adversarial networks are of high quality, excluding some cases that the scene include rivers. Experiment results suggest that such an approach is effective in images generation for infrared dataset augmentation.