Rebirth: I Carry A Desktop Computer Inside Me - #91 - Chapter 91 Research On Image Subject Analysis And Feature Recognition Methods Based On Machine Visio
- Home
- Rebirth: I Carry A Desktop Computer Inside Me
- #91 - Chapter 91 Research On Image Subject Analysis And Feature Recognition Methods Based On Machine Visio
November 11th.
The Double 11 shopping craze organized by Ali has gradually gained momentum.
But for Jiang Cheng, he couldn’t feel it.
It’s just that Yu Xinran spent more time buying some household items.
The virtuous attribute is UP.
Jiang Cheng went to his tutor Luo Ruihua and began to take the reading assessment.
There are still several questions, all in English.
But it is no longer as abnormal as the last time, from shallow to deep, from vectorization to neural networks, gradient descent and other parts of the question.
The first four questions were not difficult. Jiang Cheng had already passed the knowledge three times and could basically answer them in seconds.
The last one is to expand the extended content.
The following are the loss functions of several neural networks, allowing us to analyze the pros and cons of several types of loss functions and think about improvements.
(θ)=−1m∑i=1m[y(i) log(hθ(x(i)))+(1−y(i)) log(1−hθ(x(i)))]+λ2m∑ j=1nθ2j
…
This is not something Jiang Cheng can answer directly after reading a book, but requires his own thinking.
An hour later, Jiang Cheng handed over a neatly written answer sheet to Professor Luo Ruihua.
“Yeah, I have a good grasp of it.”
Luo Ruihua nodded and encouraged.
As for the last question, I just looked at it and didn’t comment in detail.
“Do you have any projects you plan to research? Or is it a direction that you are interested in?”
Jiang Cheng thought about it for a few days and already had a goal. He said immediately, “Teacher, I currently have two directions. One is image subject analysis and detection and classification based on machine vision; the other is efficient methods and circuit design for deep learning. .”
Luo Ruihua was a little surprised. He went in two directions at once. He was very brave. Can he bear it?
“First talk about your considerations on these two options.”
“Okay, machine vision should be said to be the analysis of the main body of the image…” Jiang Cheng had some thoughts in his mind, and when he explained it in front of Professor Luo Ruihua, he was very clear and logical.
The research direction is one thing, but to put it bluntly, the first direction is intelligent classification of images.
Jiang Cheng felt it in his previous life.
It can automatically categorize the pictures on your phone for you.
Even people of different ages in the picture can be intelligently analyzed and classified for you. Basically, there are not many problems in identifying them.
Then the second direction is artificial intelligence software and hardware collaboration.
In Jiang Cheng’s thinking, today’s circuit design and logic are actually not suitable for artificial intelligence. Even with deep learning and neural network algorithms, they are still simulated to realize various functions of artificial intelligence.
The efficiency is not enough, and the error is also a troublesome problem.
Therefore, Jiang Chengcai came up with the research idea of software and hardware collaboration.
Even doing some superficial research is fine.
“The second project is too big for you to grasp.” Luo Ruihua waved his hand, “I just read the questions and you still have ideas on machine vision. Let’s study the first one first.”
“OK.”
Later, Luo Ruihua talked about some of the cutting-edge foreign research directions in machine learning and neural networks, as well as Jiang Cheng’s specific research methods for selecting topics, including how to apply for projects and funds, and how to find a team to work on together.
“…Scientific research projects cannot be done by one person alone. You must learn to allocate tasks reasonably and scientifically coordinate the progress. I will be the instructor when applying for the project, but I will only mention a few points. The key is for you to explore on your own.”
“I understand.” Jiang Cheng kept these instructions in mind.
This is the time for the big guys to impart knowledge.
It’s not long, just over an hour.
But Jiang Cheng felt that this was much more effective than learning and exploring knowledge on his own.
Especially after his extensive research and study, he suddenly felt enlightened and enlightened in this research direction.
No wonder there is a saying that listening to you is worth ten years of reading.
—(^o^)/—-
Jiang Cheng sorted out the project application materials the next day and submitted them to the department.
Project name: Research on image subject analysis and feature recognition methods based on machine vision
Project Lead Supervisor: Professor Luo Ruihua
Specific person in charge of the project: Jiang Cheng, 11th grade undergraduate
Project application capital: 500,000 yuan.
…
Although Professor Luo has the reputation, this kind of scientific research project will not be completed so quickly.
Moreover, the results of scientific research theoretically belong to the school.
However, in recent years, in order to mobilize the enthusiasm of scientific researchers, project units are generally given patent autonomy, which can be implemented, licensed to others to implement, transferred, priced into shares, etc.
Therefore, Professor Yang Mingde has the power to decide (or basically decide) the transfer of the patent for facial recognition.
After finishing his school work, Jiang Cheng took the time to visit Fenghuo Network Technology Company.
The company has grown to more than 50 people.
Several important departments have more than five or six people.
In the planning department, Jiang Cheng handed over to Chi Fei, who took the lead in planning “Happy Xiaoxiaole”.
At present, the development process is already being accelerated.
Several casual games are still operating normally.
Monthly turnover is stable at around 20 million yuan.
Inside the office.
Although Jiang Cheng came here less often due to his busy schedule at school, the office was still cleaned every day, so it was not covered in dust.
“…The game engine development department has recruited 6 people. According to your wishes, they are all familiar with the Maplefire PE special effects software for the time being.” Chen Feng reported on the work situation, “There are 7 people in the PC game department, and they are currently developing Collapse World. “
After listening to the report, Jiang Cheng nodded and said, “Follow the established plan and develop steadily.”
“Okay, Mr. Jiang, with your planning guidance and technical support, there will be no problem in any game our company makes.” Chen Feng smiled.
The game companies I worked for in the past, the board of directors or the boss, didn’t understand anything. They wanted to add advertising and recharge. They were all about making quick money. They didn’t have Jiang Cheng’s philosophy of making games as carefully as he did.
“I plan to establish the Maple Fire Group Company based on the Maple Fire companies. At that time, you will be the general manager of Maple Fire Network Technology Company. You should work hard on the several games on hand.” Jiang Cheng reminded again, ” In addition, I plan to establish a dedicated Maplefire user community to unify the accounts and passwords of Maplefire’s various game platforms, online articles and other platforms. When the time comes, you will have to connect them well and transfer the player information of the game platform in place.”
Chen Feng said, “It’s time to unify user accounts. Otherwise, just relying on linking Weixin and deducting accounts will easily be restricted.”
After giving some instructions on some matters at hand, Jiang Cheng left the office and went to the employee work area to wander around.
After the venue was expanded, the entire functional area also changed.
Departments such as art, mobile games, client games, and planning have all designated areas.
There are more handsome guys and beauties in the art department, and they are all currently designing the original paintings of Valkyrie.
Jiang Cheng developed Honkai Impact 3 as a PC game based on the memory of his previous life (crossed out, as requested by some readers).
Each beautiful girl is painted with her own characteristics.
If Jiang Cheng was asked to paint, he probably wouldn’t be able to do it.
One or two is okay, but if there are more, Jiang Cheng himself won’t see many women, so he really can’t draw new tricks.
Just as he was about to leave, a loud quarrel came over.
“What do you mean? You say no here, no no there. If you can do it, go ahead. Being the person in charge is great, isn’t it?!”