Augmentation of Images through DCGANs
Himanshu Arora, Samyak Jain, Sanket Anand, Dharmveer Singh Rajpoot · 2019
Now-a-days, extending images has become a challenging task to implement. Many of algorithms like convolution neural networks (CNN), Generative Adversarial Network (GAN) are used to fill out the image spaces. But the challenge arrives to guess or make appropriate assumption of image extended borders. We used Deep Convolution Generative Adversarial Network(DCGAN) over GAN and CNN to implement it.[1], [2] GAN's are implemented to guess the space in image by generating fake examples using generator and decides using discriminator which determines how much the image is real. DCGAN is improved version of GAN which generates spatial correlations [5].