GAN_CNT

Generation of tangible structures using deep learning technique: conditional generative adversarial network (GAN). The four models of conditional GAN were trained by scanning electron microscope (SEM) images of carbon nanotubes with different magnifications. Trained models stored in this file can generate virtual structure of carbon nanotube at arbitrary composition.基于深度学习技术的实体结构生成方法：条件生成对抗网络（Conditional Generative Adversarial Network，GAN）。本数据集采用不同放大倍率下的碳纳米管扫描电子显微镜（SEM）图像，对四款条件生成对抗网络模型完成训练。本文件所存储的已训练模型，可生成任意组分的碳纳米管虚拟结构。