Font
Large
Medium
Small
Night
Prev Index    Favorite NextPage

Chapter 939 Image Recognition(1/2)

It’s 2008 and the economic crisis is breaking out.

It's time for something big to happen.

Zhou Buqi’s strategy for Twitter has completed its third transformation.

For the first time, when I first became a shareholder, I tried my best to help and sent Liu Qing to balance the relationship between the founder and the management, so that the smoky company atmosphere could improve and the hippie CEO could restrain himself.

This is the friendliest time.

In the second stage, starting from the second half of last year, the attitude towards Twitter has changed. Twitter has made money in its B-round financing and wants to expand overseas markets, which is not in line with Ziweixing's plan.

This coincides with Ziweixing's strategy.

Zhou Buqi entrusted Liu Qing to silently hold back the company and try every possible means to delay Twitter's overseas development plan. The result was that Twitter went to Europe instead of Asia.

The third stage is now.

There is no need to hide and tuck it away, you have to face it head-on!

Twitter must be firmly limited to the US market, and overseas markets must be conquered by yikyak. It will be launched in Japan in February, Weibo will be launched in China in March, and yikyik will enter the European market in June-July.

At this stage, conflicts are very serious.

There might be a fight on the board of directors.

Liu Qing will definitely die.

She got married last year and is pregnant. Her husband is from China and she wants to raise a baby. After giving birth to the child, she also intends to return to China to develop her career in order to avoid the feeling of disharmony between the couple who are separated in the two places.

This requires that there must be a heavyweight person sitting in Ziweixing's global market.

Lu Qi is an important candidate.

He is one of the highest-ranking Chinese in Silicon Valley IT companies.

Lu Qi is Shen Xiangyang's junior fellow student, and they have a very good relationship. In his previous life, Shen Xiangyang worked at Microsoft, and it was he who suggested to Ballmer that after the failure of the Yahoo merger, he hired Lu Qi and became a senior vice president at Microsoft.

.

The next day, Zhou Buqi signed the CEO document, promoting Zhang Hongguang, executive vice president of Ziweixing Research Institute, and Xu Ming, vice president, to M6 at the same time to maintain the balance between the engineering academy and the research institute.

The position of deputy dean is a management position related to technical research.

It is different from Yang Zhenkun’s high-level technical positions.

No matter how high his rank is, Yang Zhenkun cannot be a CTO or be in charge of a piece of business. He is a pure scientist. However, as the vice president of the Research Institute and the Academy of Engineering, he can become a candidate for CTO in the future. He can also be transferred to work in other business groups.

President or Vice President.

As soon as the promotion order came out, Shen Xiangyang came to report with a rosy face.

Zhou Buqi thought it was Lu Qi who had some clues.

"How is that possible?" Shen Xiangyang couldn't laugh or cry. "This matter cannot be rushed. It depends on the merger of Microsoft and Yahoo, as well as the pressure exerted by the board of directors on the management during the economic crisis and stock price collapse. Let's wait a few months. What I said

It's another thing that you must be paying close attention to."

"What's up?"

"A task you assigned me last year."

"Yeah?"

Zhou Buqi is a little embarrassed, he forgot about it a long time ago.

Shen Xiangyang said: "There are a lot of pornographic pictures on the Internet, almost every forum has them, and they all need a manual review team. You just ask the research institute to develop a set of algorithms that can automatically identify pornographic pictures."

Zhou Buqi immediately remembered, "Yes! You also said that you would set up a special position for pornographic appraisers, saying that women are more suitable than men."

Shen Xiangyang laughed.

Boss Zhou’s focus has always been so different.

"The role of the pornographic appraiser is to review some overly complex pornographic images and then develop algorithm models. This is a relatively deep level and has not been studied yet. My team has developed a universal pornographic image recognition The algorithm can identify more than 95% of pornographic images. As long as they are not hidden too deeply, they can be found."

In the field of computer vision recognition, Shen Xiangyang is definitely one of the best scientists in the world.

When he was studying for a Ph.D., his supervisor was the famous Professor Roger Reddy. This old Indian man specialized in speech recognition. However, Shen Xiangyang took an unconventional path and started the research on visual recognition despite the pressure. A great success.

Even ten years later, many basic theories in the field of computer vision recognition were developed during Shen Xiangyang's Ph.D.

Zhou Buqi's eyes lit up, "So fast? Did you make it?"

Shen Xiangyang smiled and said: "The recognition rate is 95%, and the accuracy rate is also about 97%. I originally thought that I could further optimize it and increase the accuracy rate to at least 99.9%. But something happened now, and I think you need it very much. The accuracy rate of 97% is enough. It’s better to market it.”

"well!"

Zhou Buqi was very excited and praised him twice.

When doing scientific research, you cannot develop without developing.

Scientific and technological achievements must be tested in practice in the market.

This small project of Ziweixing Research Institute seems to be successful! It is very likely to make a lot of money!

Automatic algorithm recognition of pornographic images eliminates the need for manual review.

AI can be used to replace labor, and there is no need to pay wages, benefits, and five social insurances and one housing fund, which can greatly save expenses.

More importantly, Shen Xiangyang was right, something happened recently.

The photo scandal went viral.

People all over the country know that the entertainment industry in Hong Kong and Taiwan is such a mess! It turns out that the glamorous and aloof female stars are the same even if they take off their clothes. They also have to pose and pose, and they have to kneel down.

In the past few years, there have been various scandals, and there have always been large-scale pictures uploaded.

But this time it was different, it involved too many goddess-level celebrities, and the popularity was unprecedented.

The national regulatory authorities have taken a heavy hand. Instead of making phone calls, they have directly issued stamped red-headed documents, requiring Internet companies in various places to purify the online market and provide young people with a healthy and green growth environment.

If you don't do it well, you will be fined or even shut down.

However, for many websites with high traffic, tens of thousands of posts are posted in a short period of time. How many image reviewers are needed? They can review during the day, but what should they do in the early morning?

This is the most troublesome thing in the industry recently.

Even the school forum exploded.

Private photos of actresses are posted and deleted again and again by college students, often making them very irritable.

It's better now.

AI is coming!

This can not only solve Ziweixing's dilemma, but also help colleagues solve their current troubles.

Zhou Buqi was very excited and asked: "How far can other people's pornographic image recognition programs go?"

Shen Xiangyang said confidently: "Even Google and Microsoft will not surpass us. In the domestic and even Asian markets, our image recognition program can represent the highest level of artificial intelligence."

Image recognition is not easy, it is even more difficult than speech recognition.

Many pictures are too complicated.

Once the accuracy rate is not high, something will happen. If a user posts a little yellow duck, a picture of a bra, or a selfie, it may be judged as a pornographic picture and cannot be posted.

This will greatly affect the user experience and will be scolded by users.

Therefore, pornographic image recognition algorithms have been around for many years, but because the accuracy rate is not high, they have not been widely accepted by Internet companies.

A very important part of the reason is that Silicon Valley has not done deep research in this field. The United States does not care about a harmonious society, so pornographic pictures are just pornographic pictures. Post them casually. Pornographic pictures are everywhere on Facebook and Myspace. Maybe they can attract traffic.

If Silicon Valley doesn't do it, the technology in this field won't be updated and it won't be used domestically.

Fortunately, Shen Xiangyang is there.

"Okay, Old Shen, you have helped your colleagues solve an urgent problem." Zhou Buqi was very happy and asked tentatively, "Well... I think this is it, transfer this program to the Academy of Engineering, and the Academy of Engineering will coordinate and outsource it. What do you think about the sales and after-sales service?"

Shen Xiangyang nodded, "That's right. It's okay. I'll communicate with the institute."

Zhou Buqi was very satisfied and said with a smile: "Don't worry, we are going to sell it at a good price. I will send you a big red envelope."

Shen Xiangyang said: "Not only domestically, you can also sell in South Korea and Japan. Especially in Japan, they also have relatively strict legal regulations and must have codes. Japan's Internet technology is much worse than that in China. Well, at least it is better than ours. MSI is much different. It is necessary to establish such a sales system in Japan, or directly entrust Yahoo Japan to sell it as an agent."

Zhou Buqi took it seriously and said, "Okay, let's discuss it later."

Many Internet companies have successful models and have little technology. When they encounter complex technical problems, they have to find foreign aid. For example, Alibaba currently relies on the technical support of Microsoft and Oracle.

Zhou Buqi is also actively working hard to become Alibaba's largest technical backend.

Japanese websites are not as good as Alibaba. They are very rough. It is a bit like the domestic banking system. Simple technical problems are handled by yourself, and slightly more complicated ones are outsourced.

It is indeed necessary to develop.

Zhou Buqi called Cheng Binghao over to discuss the matter.

Cheng Binghao shook his head and said: "I just came back from Japan the day before yesterday. After inspection, I still can't do an engineering institute there. Those Japanese people are too grindy. They are always grinding away at trivial matters. They are so annoying. And you don't know. The after-sales service there is deformed. Big companies like Sony and Panasonic are still doing after-sales maintenance for old equipment from 20 to 30 years ago. After-sales department for used equipment.”

Zhou Buqi frowned, "What do you think?"

Cheng Binghao said: "Just set up a good sales system in Japan, be stronger, only sell products, and don't accept demand orders. Communicating with those people is a waste of time and energy, and the cost of uncertainty is too high."

in the country, the high-end outsourcing of engineering institutes can meet demand.
To be continued...
Prev Index    Favorite NextPage