lone1
@lone1
Profile
Registered: 1 year, 4 months ago
How to distinguish the video produced by Deepfake and avoid the scam of false information? In the ten years since AlexNet participated in the ImageNet Large-scale Visual Recognition Challenge held in 2012 and achieved the lowest Top 5 error rate of 15.3%, the development of AI technology can be described as advancing by leaps and bounds. AI applications have shined in the fields of images, audio, etc., and the ability to identify data has begun to surpass that of humans. In 2014, scientist Ian J. Goodfellow developed a generative adversarial network called GAN (Generative adversarial networks) that has a huge impact on the future; this technology, as the name suggests, can generate extremely real data. Face-changing software: Deepfake With the gradual popularization of AI applications, it is becoming easier for people to obtain various AI tools and services for their own purposes, including Deepfake, the face-changing software that has become famous this year. Deepfake is a GAN-based application that uses face recognition technology to find the face position in the screen and replace the original face with an AI-generated face. After the public can use Deepfake, we will certainly see many creative applications, but negative examples also exist and have greater influence; they have seriously disrupted social order. At present, the most abused cases of Deepfake focus on pornographic content and political purposes. DeepNude is a tool that allows users to achieve nude results after removing clothing by inputting female photos. The release of DeepNude illustrates how the tools to use deepfakes for pornography are becoming increasingly accessible, and any woman can fall victim to the misuse of this technology. A great invention of scientific research has been misused to create false information and group manipulation. The coordination of AI technology and ethical norms is a process that we need to reflect on and improve. At the same time, the public also needs to cultivate awareness of new types of crimes in order to prevent them Not yet. The following are the views compiled by the MIT media lab, which will help the public other than researchers to increase their sensitivity to the false content of Deepfake. Pay special attention to faces in videos, since deepfakes are almost always used to transform faces.• Notice the subject's cheeks and forehead in the movie, does the skin appear too smooth or wrinkled? Is aging of the skin similar to aging of the hair and eyes? The results generated by deepfakes are often inconsistent in some ways.• Pay attention to the eyes and eyebrows, do the highlights and shadows match the look of the scene? Deepfakes are often unable to fully and perfectly represent the natural physics of a scene.• Please notice whether the person wears glasses, and notice whether the glasses reflect light when the person's head moves? As in the previous case, this is because deepfakes usually cannot perfectly represent the natural physical characteristics of lighting.• Notice the facial hair, does it look real? Deepfakes may add or remove mustaches, sideburns, or mustaches. But deepfakes often fail to make facial hair transformations completely natural.• Pay attention to the size and color of your lips. Does the size and color match the rest of the face? Jevin West and Carl Bergstrom, professors at the University of Washington in the United States, set up " WhichFaceIsReal.com ". There will be two photos of human faces on the website, and you will be asked to choose which one is the real face. In the process of selection, the ability to judge fake faces can be gradually cultivated.
Website: https://techtuba.com/
Forums
Topics Started: 0
Replies Created: 0
Forum Role: Participant