Help all mankind

Professor Flanker passed a picture as soon as the voice fell.

It is a black and white dot matrix picture. The special thing is that there is a cartoon character on the thumbnail, but after clicking it, it shows a building.

Liang Shicheng knew the principle and typed it disapprovingly,

[Professor Flanker, this picture can only fool the computer screen.

It's just because of the problem of the selected pixels when the thumbnail is displayed on the screen, which causes the picture to be different at different zoom ratios, which is completely different from the pattern of the truck box.

Professor Flanker responded,

[Well, you can only deceive the screen, but have you ever wondered if the human eye has limitations and can be deceived?

This is just an example, I want to give you a general impression so that you can think in this direction.

Boy, listen, the next thing is the point.

Later, Professor Flanker passed some special pictures over.

Some of these pictures are still, but as long as the eyes are swept over the picture a few times, the surroundings of the picture seem to start to move, but once the eyes are fixed on a certain position in the picture, the position becomes static and no longer moves. .

Some pictures are originally arranged neatly with some patterns, but they always give people a messy illusion. Even the straight straight lines in the patterns seem to be curved when you look at them. You can only find out if you look very carefully. They were originally straight lines.

The last picture is a moving picture, with notes underneath the picture, staring carefully at the middle, the surrounding images will disappear.

After seeing this picture, Liang Shicheng did the same and stared at the middle. As expected, after about ten seconds, all the surrounding images disappeared, but the attention of his eyes was slightly moved to the side, and the disappeared patterns returned. Up.

Although these pictures are very strange, Liang Shicheng is not surprised.

Liang Shicheng is not unfamiliar with these kinds of pictures, and it is not the first time he has encountered them. He just thinks that these pictures are so strange because of some psychological reasons.

After reading the picture, Liang Shicheng replied in the chat software,

[Aren’t these just illusion pictures?I have been exposed to these pictures before, and I have an impression of them. What do these pictures have to do with the disappearing truck?

Professor Flanker did not reply immediately. It seems that his typing speed is very slow, but that is not the case. At the moment, the professor's computer is full of various materials.

The professor sorted out these materials patiently so that Liang Shicheng could understand, and it took a while to explain.

[I want to show you these pictures to tell you that seeing is believing in this world.

Human eyes are easy to be deceived. Human eyes cannot see the whole world. What we see is only the surface of the world.

Immediately afterwards, Professor Flanker sent an upside-down human face. At first glance, it seemed that there was no problem at all. The professor asked Liang Shicheng.

[Do you think there is something wrong with this picture?

Liang Shicheng felt that everything in the picture was normal, and replied,

[It seems to be a very ordinary photo, no problem was found.

Professor Flanker was a little proud, but he didn't show it in the words, he just replied indifferently,

[Then turn the picture 180 degrees.

After Liang Shicheng turned the picture 180 degrees, he felt very strange.

Professor Flanker posted many similar pictures one after another. There are many problems in the careful analysis of these pictures, but there is no sense of disobedience when viewed with eyes, which made Liang Shicheng feel a little confused for a while.

When Liang Shicheng was still worried, he suddenly received a new reply from the professor and pulled him back to reality.

[Okay, let's go back to the previous topic, why does a crash happen.

After giving an example for a long time, Professor Flanker finally started to go straight to the topic.

[Let me put it bluntly, there is such a picture in this world that looks like a human eye, looks like a different animal, and looks like another in the eyes of artificial intelligence.

The human eye cannot directly see the original complete information of a picture. It should be known that a picture generally consists of millions of pixels.

In order to reduce the burden, the human eye only extracts the superficial information in the picture and generates a false image in the brain.

In other words, in the end, what we see in our eyes is only the most superficial things in a picture, and what is buried in this image is invisible to our human eyes.

However, the surface and inner layers are just a metaphor, not stacked one after another.

This is a little bit mysterious, but you just need to know what it means.

Professor Flanker's introduction made Liang Shicheng amazed, although he has little knowledge of biology and psychology, and about human eye imaging.

But as an experienced engineer who studies artificial intelligence and image recognition, he can immediately understand what the professor means.

But he continued to ask modestly,

[Surface image?Inner image?Can you give some more examples?

Professor Flanker thought for a while and continued to describe it tirelessly.

[That's it, you give me a picture now, I can use a special method to process this picture.

The new picture after processing does not seem to change in your eyes, but I have added other images in this picture, but the added image is placed in the inner image, and only the eyes of artificial intelligence can see clearly.

In the inner image world, no matter what picture you give me, it is a completely blank canvas for me. I can paint at will, but no matter what I paint, the human eye cannot see it. But artificial intelligence can see.

[I said so, can you understand?

Then Professor Flanker sent a website, which is a research report on the official website of the University of Chicago, which has a detailed introduction to this topic.

Many scientists are already studying this field.

Although it was some more rigorous and formal research reports, Liang Shicheng as an engineer quickly understood the general meaning, but before he could finish reading these contents, Professor Flanker gave some amazing insights.

[Can you now understand why your car hit that truck?

Because the pattern on the truck box itself is exactly the same as the background blue sky and white clouds, in the eyes of the driverless system, the truck easily blends with the background, so a car accident occurred.

But that pattern has another look in people’s eyes...]

Liang Shicheng suddenly realized, and immediately replied,

I understand!You mean that the pattern on the car box deceived the artificial intelligence picture recognition system!

[No, you are wrong, I mean, the pattern on the car box deceives human eyes.

Different eyes see the world differently, human eyes, animal eyes, and artificial intelligence eyes.

Tiger markings are so obvious in human eyes, but they can be invisible in animal eyes. It is only because animal eyes lack the perception of color. Human eyes can see much more than it, which is equivalent to seeing The inner world of animals is closer to the real world as seen by human eyes, but it is still not a complete real world.

Just like the pictures I just showed you that can produce hallucinations, they can't fool the artificial intelligence, so I prefer artificial intelligence to have a more realistic perspective.

From this perspective, it is still the human eye that is deceived.

To be precise, the real pattern on the car box is the blue sky camouflage. The surface of my analysis software is 98% similar to the real blue sky, but after some processing, the human eye has an illusion. Another image.

PS superimposes other pictures in a picture, and eventually the human eye sees a picture, and artificial intelligence sees another picture.

This is not the author's whimsical imagination, but the technology in reality. The reality may be more evil than science fiction and the author's imagination.

The inspiration came from a foreign paper, and the relevant content was also found on Zhihu in China

address

Next, use this technology to deliberately produce some pictures that can deceive the human eye. In the eyes of artificial intelligence, it is another picture. The technical battle triggered by this is also not the author's own ideas, but the reality of drawing materials.

I wrote some technical things again. It took time to check the information. I didn’t send it until 11 o’clock. It was thankless. You can rest assured that I will write pure plot content in the second volume.

In addition, when I wrote this chapter, I realized that it is really troublesome not to post pictures. In fact, this chapter should be matched with a large number of pictures. I will try to upload the pictures to the comment area after I try to upload them, or also You can upload pictures to the group.