Rebirth: I carry a desktop computer inside me

The overall project plan is divided into three steps.

The preliminary work, mainly 2 items, based on the project direction, preliminarily proposed the algorithm of image subject recognition, and collected the image library to ensure the follow-up experiment.

In the medium term, it is to experiment and train the recognition algorithm, during which it is necessary to constantly modify and improve the algorithm and continuously enrich the database through machine learning.

In the later stage, organize the experimental data, improve the algorithm, and integrate the packaging into a paper, and if possible, apply for a patent.

However, after the strong upgrade of the computer in the entity, the 6-core and 12-thread virtual machine vacated by Jiang Cheng was already idle after the special effects software was updated to version 2.0.

After declaring the project, Jiang Cheng's virtual machine began to write code like crazy again.

As we all know, machine vision is not the same as our human vision.

The beauty on a picture, we all know with our eyes that this is a beauty.

But machines are different.

In the eyes of the machine, this picture, whether it is a beauty or a beast, is a string of 0s and 1s.

Therefore, machines need to simulate our human brain for recognition.

Convolutional neural network CNNs are one of the excellent and extensive solutions.

Divide a complete picture into multiple small parts, extract the features in each small part, and then summarize the features of these small parts together, so that the process of machine recognition of the image is completed.

But there are a lot of problems here.

For example, each small piece, many features near the pixel are very close, if all the computing power is put here, it is a waste of computing power.

And it increases the burden of identification.

For example, at the beginning, even if some features are recognized, the machine still does not know what these features represent.

These problems are all that Jiang Cheng needs to solve in this project experiment.

Of course, some problems, in the current research process, there are already some solutions at the frontier, especially in foreign frontiers, each with advantages and disadvantages.

The next day, at a table in Guanyuan Garden, Jiang Cheng took his laptop and prepared to discuss and arrange the project content with Wang Sichen.

"For our part, we need to use the latest Python 3.1, the environment configuration plus tensorflow and opencv, and the compiler uses Spyder." Jiang Cheng was on the computer screen and opened the compiler.

Wang Sichen was a little confused, although he also had contact with the Python language, he was not very familiar, and he was already making up and training last night.

But he didn't expect that the next day, Jiang Chengcheng began to discuss the algorithm implementation plan.

Jiang Cheng glanced at it, knowing that he could not count on discussing this aspect, and directly ordered, "Okay, I will come to the specific algorithm myself, and I will talk to you in detail when I have the opportunity, you will go to the Internet this month to download character pictures, all kinds of backgrounds, regardless of men, women and children."

"Okay, crawler in Python?"

"Yes, of course, you know the restrictions on use, and you have to abide by the robots protocol." Jiang Cheng nodded and handed him a 1T portable hard drive.

Wang Sichen took the task, "Teacher Jiang, don't worry, I promise to complete the task." Between

meals, Jiang Cheng asked something, and probably knew about Wang Sichen's situation.

A native of Huizhou Province, his high school grades are also quite good, but he has not been admitted to a very good university.

Because I heard that the computer major is good and has great development prospects, I chose the computer major.

Then I stared at Mizuki for graduate school and wanted to go to the best school.

The personality is a little introverted and not good at words.

But people are honest, when they knew that Jiang Cheng was a freshman, he published many papers, participated in the big guy experiment, and now he is a direct disciple of Professor Luo Ruihua in advance, and he is even more admired.

In speech and behavior, I still respect each other.

This is good, Jiang Cheng thought it would be better than giving himself a prick or an assistant who was too level.

Although Jiang Cheng is not afraid of thorns.

"Hurry, my research speed is relatively fast, you have to adapt to the rhythm as soon as possible." Finally, Jiang Cheng reminded.

Unlike some who procrastinate for project funds, his most important thing is time.

What he is most afraid of is that his task will be completed quickly in the internal computer, but he still has to wait for a large number of pictures to conduct training experiments.

----o(-"-)o-----

December 24, is the time for public auction of several plots of land in the Kyoto Economic Development Zone.

Jiang Cheng specially came over to take a look together.

On hand, the funds are ready.

The group has its own funds of 40 million yuan and loans of 30 million yuan, a total of 70 million yuan.

The starting price is 52 million yuan, if the premium is too high, it is not cost-effective.

After all, it is not commercial or residential land, the premium is too high, and the cost of corporate land is too large.

And this is also 12 years, and a few years later, more than 100 acres of industrial land, at least 200 million to get it.

"Mr. Jiang, our auction number is M1650, and a total of 4 companies will participate in the auction this time." When they arrived at the place, Guo Xiaoming and Jiang Cheng reported the situation.

4 companies, including Oriental Muge, Shunyi Automobile and other enterprises.

But judging from the information Jiang Cheng found on the Internet, Dongfang Muge had some ideas and strength.

The other two, Jiang Chengdu is estimated to be suspected of making up the numbers.

In Beijing in 2012, the transaction volume of industrial land was somewhat lower than last year, and the market investment enthusiasm was also somewhat reduced.

Many people are crowding towards commercial and residential land.

This is a lot of money!

Spend a lot of money to take land, after development, and then sell houses with high quality and high prices, the house price is clearly watched, it will definitely rise, of course, they are also pushing up.

And Dongfang Muge, as an air conditioner, kitchen and bathroom type enterprise, spent so much money to buy such a large piece of land in Kyoto, Jiang Cheng felt that it was unlikely.

The two sat down at the auction center.

Soon, after the auction began, the host began a passionate auction presentation.

This time, there are three parcels of land, one commercial and residential, and two industrial land.

The commercial and residential land has been auctioned, but the transaction price of 40 acres of land has exceeded 1 billion.

And the next land is the plot that Jiang Cheng fancy.

"This land is the N17B2 plot of Jinghai Road, with a construction land area of 91,512.4 square meters, a square meter plot ratio of 2.5, and a construction area of 228781 square meters... The starting price is 52.806 million yuan, and each price increase shall not be less than 100,000 yuan, not higher than 3 million yuan, and the auction will start below!

"Good, M1648 raised a plaque of 55 million yuan."

Only for the first time, representatives of Shunyi Auto raised the price.

And the increase is high.

The scene waited for a while, and when the host shouted, "55 million yuan, the first time", Jiang Cheng signaled Guo Xiaoming to hold up the sign.

"55.1 million yuan."

Each time, only 100,000 yuan is added.

"Well, M1650 raised a plaque of 55.1 million yuan. Is there a higher price? The

representative of Dongfang Muge raised a card.

"M1646 raised a plaque of 55.5 million yuan."

............

At the beginning, everyone was just trying to raise the price.

After a while, the price rose to more than 58 million yuan.

The premium is already almost 10%.

A few of them had already given up the bidding, leaving only Dongfang Muge and Jiang Cheng's Fenghuo Group.

The bidding has also exceeded 60 million yuan.

At this price, it is already very uneconomical to go up further.

For enterprises, it is to pay attention to cost control.

At Dongfang Muge, Deputy General Manager Zhang Dongfang was also a little anxious.

"M1646 raised a sign of 60 million yuan."

"M1650 raised a plaque of 60.1 million yuan."

"60.4 million yuan."

"M1650 raised a plaque of 60.5 million yuan."

......

"Which company is this Maple Fire Group, and why did it suddenly appear."

Assistant Xiao Qin did some homework, "It has only been established for a few months after checking, but there is a news circulating on the Internet that Warwick and Maple Fire Group have cooperated." "

What for?"

"Semiconductors, games, whatever."

"Add 2 million yuan directly, and then forget it." Dongfang Muge wants the land, but it is impossible to fight with the other party in order to get the land.

In case you add it, people don't want it.

Fuck, only add 100,000 yuan each time, it's very upsetting!

This makes people have to consider, could it be a lure?

It seems that I only need to add a little bit every time, but in fact, the cost is slowly rising.

"62.5 million yuan at a time." The host is also a little depressed, the two companies have been fighting, and they are both very slow, making people have to talk a lot.

I couldn't drink the water again, and my mouth was dry.

"62.6 million yuan."

Jiang Cheng continued to raise the price by 100,000 yuan.

Dongfang Zhang of Dongfang Muge was simply crazy.

Add it, the other party has always been a little higher than you; Don't add it, it's only 100,000 yuan, and there is always a little itch.

"Forget it, don't want it." Zhang Dongfang waved his hand.

"M1650 raised a plaque for 62.6 million yuan once."

"62.6 million yuan, twice."

"62.6 million yuan, three times, deal!"