"How much? 99% and 1 second? Zheng Shijian was desperate when he heard this data.

The accuracy rate of the algorithm made by Hanwang Company cannot even reach 50%, and it also requires people to be looking at the camera all the time, and the recognition rate is also 5 to 10 seconds.

Compared with Ma Tian, it is not an exaggeration to say that it is rubbish.

From 03 years of hard work to the present, the results of 8 years are all rubbish!

Zheng Shi felt that his spirit was drained all of a sudden, he stiffened, and what he wanted to say finally became: "Thank you Mr. Ma for informing me, I don't have any problems, I'm bothered today!" ”

After saying the weak words, Zheng Shi saw that he got up and was about to leave.

Ma Tian watched all this, but his eyes were flashing.

He thought for a while, and spoke out to stop Zheng Shijian from leaving: "Don't be in a hurry to leave, tell me about the current development of face recognition technology at home and abroad!" ”

Zheng Shijian looked at Ma Tian, but returned to his original position and said:

In 2001, Dr. Shasuha proposed the face recognition and drawing technology based on commercial images, and the definition of face signature images with different lighting conditions can be used for face recognition under the condition of constant illumination, so that people are aware of the research of face recognition light image algorithms.

Later, Dr. Basley and Dr. Jacob used spherical harmonics to represent illumination, and used the convolution process to describe the reflection to prove that the set of all Lambo reflection functions obtained by any far-point light source forms a linear space! Their proof provides a way to solve the problem of face lighting transformation.

Now, basically, we continue to optimize the accuracy of face recognition under different lighting based on these two bases.

At present, the world's leading technologies are Cognitec and Idetiix, which can achieve an accuracy of 80% based on rectangular features and AdaBoost feature detection! ”

Zheng Shi saw that he was eloquent and obviously had a deep understanding of the international research on face recognition technology.

Ma Tian nodded with satisfaction: "What about what your Hanwang company researched?" ”

Zheng Shijian: ......

Zheng Shi was silent when he saw it, as the so-called beating people does not slap their faces, what is the difference between Ma Tian's actions and slapping his face.

However, he still replied truthfully: "Now we have relatively little research on face recognition algorithms, and it is not yet live detection, and the accuracy rate is about 30%!" ”

This time it was Ma Tian's turn to be silent, he didn't expect such a dish in China. In addition to the fact that he is a little behind foreign countries, everyone has uploaded to the point where the algorithm has landed, and China is still just starting to research.

Zheng Shijian saw that Ma Tian was silent, and finally couldn't help but ask his doubts: "Mr. Ma, I want to ask, how did you cross so many technologies at once and achieve a 99% accuracy rate?" ”

It's not that he doesn't believe in Ma Tian, but it's really a bit outrageous. Technology is to be accumulated from generation to generation, how can it be so easy to climb to the sky in one step.

He didn't know that Ma Tian was a hanger, and the mathematics of Lv3 had almost sent the final version of the face recognition algorithm to Ma Tian's mind, and Ma Tian could write it by looking at the face picture a little.

"Because the technology I'm using is deep learning, or neural networks!" Ma Tian replied.

Deep learning is a complex machine learning algorithm that far surpasses previous technologies in terms of language and image recognition.

In short, it's like a neural network given to a robot-like class with the ability to learn and summarize. For example, for a face image, it can allow the machine to divide it into three layers like a human to speed up the recognition.

When we normal people recognize a face, we first look at the facial features, right, then we may look at the hair and skin wrinkles, and finally judge whether it is a face by the body below the head.

Machines that use multi-layer neural networks for deep learning can also roughly have this ability.

"Are you talking about machine learning guru and Toronto professor Kilfri published in Science in 2006?" Zheng Shijian asked.

Ma Tian nodded, he didn't expect that the concept of machine learning had been proposed abroad, but before Zheng Shi met, he must pretend to know about this paper.

But Zheng Shijian knows a little more, and it seems that he does have a deep research on face recognition technology. Ma Tian thought to himself.

Zheng Shijian was silent, it turned out that the problem was here. In the final analysis, they Hanwang has studied less, and so many foreign technical publications have not been carefully studied.

Thinking of this, Zheng Shi couldn't help but look at Ma Tian gratefully, at least Ma Tian pointed out a correct path to catch up.

"Thank you Mr. Ma for informing us, if it weren't for your reminder, we would have to take many detours!" Zheng Shijian said sincerely.

Ma Tian smiled and didn't say much, it was not that he looked down on Zheng Shi when he saw them, but that the foreign founders were just starting out in machine learning now.

He didn't think that his prompt today, Zheng Shi would be able to make a big breakthrough when he saw it.

Seeing Zheng Shijian leave excitedly, Ma Tian also showed a meaningful smile.

He thinks that Zheng Shijian is a talent, and he is now selling face recognition doors without a R&D manager, and he thinks Zheng Shijian is a good candidate.

Now let Zheng Shijian go back and hit a wall, and then recruit to sell face recognition automatic doors and engage in machine learning, Ma Tian thinks this idea is very good.

The field of artificial intelligence algorithms requires long-term research and accumulation, and it is definitely unrealistic to rely on him alone, after all, just knocking on code and building wheels is a huge amount of work.

Zheng Shi saw that it was indeed as Ma Tian expected, and he was so excited that he was about to fly back directly by plane.

When he returned to Hanwang Company, he organized his own team meeting non-stop.

"I'm going to find Ma Tian!" He sat at the top of the house, and the first sentence attracted the attention of the audience.

"Ma Tianzhen has developed our ideal product, his accuracy rate is 99%, and the recognition speed is about 1 second!" Zheng Shi didn't stop talking and continued.

As soon as this data came out, the reaction of the people below was exactly the same as when he heard Ma Tian's answer, and he was directly desperate.

"How is that possible? 99%, 1 second, woo-woo-woo - "Someone is in pain."

"It's over, it's all over!"

……

"Now the international leading level can't reach this data." Tang Jiefei looked incredulous, and directly got up to question, but when he thought that Zheng Shi couldn't joke and lie, he immediately put on a mask of pain, and collapsed in his seat in despair.

"Don't give up, Ma Tian told me what method he used!" Zheng Shijian said in time, a blow is enough, but he can't let his team despair.

"What method did he use?"

"Deep learning!" Zheng Shijian said the answer, seeing that many people have not been exposed to this, he explained, "This is a method proposed by machine learning master and Toronto professor Kilfri published in the journal Science in 06, and Ma Tian applied this method."

After the meeting, I will post this paper in the group, and everyone will study it carefully to see if we can also figure out one of our own face recognition algorithms. ”

(End of chapter)