Shockingly, a counterfeit factory has produced a 3-nanometer chip.

Chapter 11 First Generation of African Beauty Techniques

"First, we need to establish a facial database containing 10 images of people of different races, with Black samples accounting for 40%."

Then, this data was used to retrain the face detection model.

"Second, the pipeline for processing in the same direction signal needs to be modified." The red modification marks on the flowchart were clearly visible, almost like an operating procedure, which gave the engineers in the audience a hint of what was going on.

"We perform skin color recognition in RAW, and adopt different noise reduction parameters and detail enhancement strategies for different skin colors."

Black skin tends to lose yellow and red hues in low light, so we need to protect it early on.

"The third, and most crucial, step..."

At this point, Shen Fei paused slightly, knowing that they would be deeply surprised by this news.

"We also need to develop a skin color restoration network based on deep learning."

The meeting room fell completely silent. In 2008, deep learning was not uncommon, but it was still a cutting-edge concept in academia.

Although various models are touted as incredibly promising, there are almost no real commercial applications... Even if there are, they are basically used to scam investors' money, and the real models are all idiotic.

Transsion, a small, bankrupt company, sets such high goals.

Good news, the boss knows a bit about technology.

The bad news doesn't seem to be too much; this kind of work is actually more tiring, and some engineers who chose to join because of the salary are now feeling like giving up.

Seeing everyone's hesitant expressions, Shen Fei laughed instead: "Don't worry, everyone, this deep learning model has been developed and the current results are quite good."

Let me explain it to you in detail.

Then, to everyone's astonishment, Shen Fei opened another PowerPoint presentation and began a more detailed explanation.

"This network has 12 layers. The input is a pre-processed image, and the output is the skin color optimization result."

We added a large number of multi-skinned portraits to the dataset for fine-tuning. The network can automatically recognize ethnicity, lighting conditions, and shooting environment, and perform personalized optimization for each photo.

To further stimulate these unearthed gems, Shen Fei also brought out the real stuff and showed a portion of the code!

The structure is ingenious!

"The algorithm complexity is kept within an acceptable range. On a 300MHz ARM11 processor, it only takes 0.8 seconds to process an 800-megapixel photo."

If the resolution is reduced to 30 pixels, it only takes 0.1 seconds, which is more than enough for real-time preview.

At this moment, the engineers in the audience were completely stunned. This company, Transsion, seemed to be different from what they advertised... They really had technical strength.

This processing speed is acceptable? This is insane! Even Nokia or Motorola would have needed years to develop such a perfect algorithm.

But the telepathic message had already been taken out.

An engineer raised his hand: "Mr. Shen, the patent for this algorithm..."

"Global patent applications are already underway..." Shen Fei smiled. "Patents have been filed in China, the United States, Europe, and major African countries. Transsion will have exclusive rights to use the technology for at least the next 20 years."

Chen Hai stood up excitedly: "President Shen, based on this algorithm alone, Transsion's photography capabilities in the African market will crush all competitors."

This is simply...this is simply a..."

"A killer app," Shen Fei said for the excited Chen Hai.

Another engineer suddenly realized what was happening and tentatively asked, "Mr. Shen, this project isn't something we need to develop, as you said; it's already been completed, isn't it?"

That photo wasn't photoshopped either, but rather...

So all we need to do next is simple adaptation work.

It's so much easier talking to smart people. Shen Fei stood up and clapped his hands: "That's right, this project has been launched. The photo just now was taken with a modified Transsion T100."

But algorithms are just the starting point; adaptation is equally important. I hope everyone can work their best so we can capture the entire African market as soon as possible.

"Don't worry, boss, I guarantee I'll complete the task."

We're just wondering if we'll get to meet these big shots; we have some questions we'd like to ask them.

Some engineers who originally wanted to retire were now inspired by Transsion's hidden R&D capabilities... They also wanted to learn and take their skills to the next level.

At this moment, Shen Fei turned off the projector light and turned on the conference room lights.

"No matter how powerful your algorithms are, they still need to be implemented on hardware."

Our hardware…”

Shen Fei looked at Chen Hai, "Engineer Chen, how's the prototype project progressing?"

Chen Hai was also full of energy at this moment and immediately reported his work results: "The engineering prototype of T80 has been completed. It is dual SIM dual standby and uses MT6223D. The chip price is 3.2 US dollars. All costs can be controlled within 20 US dollars."

The T1 prototype is still being tested. The 4-card, 4-standby configuration requires some optimization of the motherboard, and the battery compartment needs to be enlarged.

The MT6223D chip only supports calls and communication, with all other functions being cut. It is positioned as an ultra-low-end chip within MediaTek, but it is perfectly adequate for the African market.

"time!"

"The T80 can begin small-batch trial production this weekend, with an initial batch of 5000 units. The T1... will take at least another two weeks."

Shen Fei glanced at the calendar: "It's too slow. I'll add 300 million to the budget for the R&D center. If we don't have enough staff, we'll hire more; if we don't have enough equipment, we'll buy more."

Ten days from now, I need to see a trial production batch of 100 units of each of the two prototypes. If there are no problems, then we can begin mass production.

"Boss, we're running out of money... money is almost gone."

The newly appointed CFO, Wang Shanghua, and the purchasing manager, Liu Fang, said in unison.

In the past two days, they have purchased a lot of materials to keep pace with the T80 and T1 projects. In order to reduce costs, all of them are purchased in large quantities.

MediaTek directly purchased 100 million MT6223D units, ultimately reducing the price by $0.2...

However, MediaTek was also worried that Transsion wouldn't be able to pay, so they demanded a deposit before shipping the goods... So now their accounts are really...

"Don't worry about the money; the company will receive an additional investment soon."

"Then there's no problem!"

……

After the meeting, Chen Hai caught up with Shen Fei, his eyes filled with admiration: "President Shen, that algorithm... it really was developed by your friend."

Shen Fei looked at him: "Engineer Chen, don't delve too deeply into some issues. You just need to know that I didn't waste those years."

The Transsion of today is not the Transsion of last year; we now have a technological moat. Our next step is to transform that into a product barrier as quickly as possible.

He patted Chen Hai on the shoulder: "The first batch of goods will be shipped to Nigeria in a month. Whether we can win this battle depends on you."

……

For the next month or so, Transsion entered a period of intense research and development.

The R&D center operates on a 24-hour shift system, while the production line operates on a three-shift system.

But all the employees had smiles on their faces, after all, Boss Shen really did pay! Overtime pay was generous, and even when working overtime at night, they would receive their pay first before starting work...

There's no need for the production line team leader to remind them; they'll work hard on their own and try their best to improve accuracy.