Why did I start this website?
Thank you all for coming to this website. Hello. When I started this website, I was just a graduate student. I'm not a big shot, but I study the field of generative models that fascinate me and get excited from time to time. Because I think generating models is doing a very romantic thing mathematically: it wants computers to learn to understand how the world we see is "drawn". Whenever a team comes up with a new generation model and the generated pictures have a more realistic effect, I get excited for them and can't wait to understand what methods they used to achieve such a breakthrough.
Recently, the Nvidia team came up with a StyleGAN model that can generate high-definition and realistic photos of faces in the right training set. When I first saw the experimental photos they showed, I was a little surprised because the effect was so incredible that I couldn't accurately tell the difference between the real photo and the fake photo. Later, I tried to learn to understand their model, and downloaded their experimental data, and then I found that the playability of the model was simply too high. Not only because it can generate high-definition facial photos, but also because its model uses a hierarchical method to control visual features, so that we can adjust the features at different levels, so as to accurately control what our output pictures look like, such as long hair, short hair or wavy curls.
However, the playability of the model alone is not enough to inspire me to write a website and take great pains to promote other people's technology. What really makes me want to write this website has something to do with the recent project in our lab. My laboratory should be the top laboratory in the field of AI in our school (Amway: the sk group of the Exchange Center of the Network Research Institute, the mentor is huge Nice and the hardware in the group is excellent, brothers and girls are welcome to join). In our computer vision group, we will receive projects to do from time to time. However, I found that the projects we received were never related to the generation of models, and basically what we had to do were computer vision-related projects such as face recognition, target detection, object recognition and video analysis, which are very useful in the application market. Of course, this is understandable and acceptable, because enterprises need to use stable and generally recognized algorithm models to do related projects, while the generation model has not been recognized in industrial applications, and most of them remain at the theoretical level.
However, I think this does not prevent us from exploring the pace of generating model applications. In my opinion, for any scientific research project, as long as its idea is attractive enough and its realization prospect is valuable enough, it is worth exploring. For example, for generating models, you can imagine seeing a little brother (sister) who is beautiful enough for you to love him or her in virtual reality, and it can also help artists depict some extreme fantasies for people to see exactly. More broadly, industries closely related to vision, such as medical treatment, construction, transportation, clothing and so on, are all fields in which generative models can be widely used. To sum up, these exciting enough ideas make us believe that the application value of the generated model is huge, which prompted me to write this website, just in the hope that the generated model can get people's attention and be willing to invest in and study it, even though I'm only showing one of the small possibilities.
Of course, back to reality, and then reflect, in fact, I did not put so much responsibility in my heart, ask myself, the significance of writing this website to me may still be more fun. But finally, I would like to remind all friends who visit this website to be vigilant. No matter how well the model is depicted, it is only false. We should learn not to sink into this illusion. It is also a little meaningful for the website to learn to adapt to such an environment as soon as possible. I hope that after a few years, our eyes may not be able to tell the true from the false, but our hearts can remain pure and bright, and we can always be responsible and responsible when shuttling between virtual reality and real life.
Life must be lived with love, happiness, and dreams.