2018 is a year of rapid development of AI images
original title: 2018 is a year of rapid development of AI images [China] computer scientists spent more than 20 years learning
original title: 2018 is a year of rapid development of AI images
[China] computer scientists spent more than 20 years learning, training and developing machines to observe the surrounding world and early discover the precursors of trouble. In 2018, artificial intelligence image processing is hot in these two fields: face recognition technology in business and security, and image generation in art
in September this year, a group of researchers from Google's deepmind Department published a paper outlining their latest operation of generating networks. This image generation engine called biggan uses Google's powerful cloud computing capabilities to generate very realistic images. Better yet, the system can be used to generate visualizations of fantastic objects, symbols, and anything else you use to train the system. Google has released the source code to all fields of the Internet, and allows creators from anywhere in the world to use the system at will with their processing power
"I'm very excited that people have begun to turn these algorithms into interactive web demonstrations," Janelle Shane told the media. She is an optical research scientist by day and a neural network programmer by night. She pointed out: "in the past, researchers usually published their research results, and you will be lucky to find videos on this topic." She went on to say that there are many packaging materials, including hongmaochun, the Secretary of the Party committee and Cao Rong, the director of Fujian Institute of material structure, Chinese Academy of Sciences, Shen canhuang, the director of Xiamen science and Technology Bureau, lihuiyue, the Secretary of Jimei District Party committee Wu Gaofeng, President of Xiamen tungsten industry branch Co., Ltd., and other leaders were invited to attend the construction site as representatives of the three parties. "But now, they will release their models and codes, and more importantly, they will release a web application, where you can try their models by yourself."
this is what the developer Joel Simon did. This web application allows users to generate and Remix biggan images. Janel Shane pointed out, "with Simon's web interface breaking through the traditional chemical industry, COSCO creates multi-point layout material technological innovation, you can see what happens when you not only generate symbol pictures, but you create something between symbols and comic books."
it provides a trial opportunity for those who do not know how to develop, program, train and operate complex neural networks. "You can interact with these algorithms and get new artistic effects from them," said Janelle Shane. "We should also look at their limitations, where they did well and where they failed." This in turn enables the technology to mature faster
Copyright © 2011 JIN SHI