Meng Fanqi and Wang Kai have communicated a lot during this time and have become quite familiar.

However, when Meng Fanqi saw this sentence, he felt quite absurd: "You, a tall and muscular man, want me to take you where? Should I take you or should you carry me?"

"Brother, I've been arranged to come find you again," Wang Kai said, sending Meng Fanqi a photo of the airport. "Just arrived at the airport, the evening flight is going to Shanghai."

"What's going on?"

"The leader doesn't trust you to come to Yan Jing, so he can only send me to Shanghai," Wang Kai said helplessly. Although he had just made a contribution, he felt that the work was not very fulfilling.

"The leader said he wants me to be by your side to fully understand this part of the code, and then explain it to everyone when we go back."

"It's the same job as last time," Meng Fanqi understood. After all, he was not from Baidu, and Li Yanhong was eager to get the government project, so he definitely needed someone to hand over the code to him.

Just looking at the last part, it seems that they are not at ease. Therefore, they sent Wang Kai, who had been communicating with him all along, to follow him and fully grasp the new part of the content developed by Meng Fanqi.

"Alright, that's fine. We'll meet at the Baidu branch near Hechuan Road tomorrow."

"No problem, bro," Wang Kai agreed readily. "If you ever need someone in the future, be sure to think of me. No matter when or where, brother will definitely work with you."

"So, you're taking me with you, is that what you mean?" Meng Fanqi finally understood.

"Of course, working for someone else doesn't have much of a future. Just get a stable salary and you won't save much money in your lifetime," Wang Kai said directly, not hiding his true thoughts.

"I don't remember where I heard this, but they say that everyone encounters at least one opportunity to change their destiny every day, but most of the time they have no idea."

"Changing your life depends on how you change. If you go out and get hit by a car, that's also changing your life," Meng Fanqi had heard another version: "The version I heard was that a person in Silicon Valley will definitely encounter an opportunity to get rich overnight in their lifetime, commonly known as the 'five ones'."

"I think you're my opportunity, bro. If you really start a business in the future, be sure to call me. I will definitely work diligently and bow out only after giving my all." After experiencing Meng Fanqi's technical impact twice, Wang Kai was completely convinced of his technical prowess.

Many times, writing an algorithm is not a matter of whether you can do it, but whether you can think of it.

In the past few days, Baidu AI Research Institute has recalled a group of technical personnel who were working outside, hoping to gradually shift the focus back. However, after a busy week, there was no meaningful progress in multi-target detection and tracking.

Although Baidu AI's business is thriving recently, there are actually quite a few internal AI technical personnel who have complaints.

To read the uncut version, go to [p𝘢wread.com].

The YOLO algorithm is very powerful, but it was not independently developed by Baidu. It was just a large-scale project that borrowed the power of Meng Fanqi.

There has indeed been rapid progress in the short term, but in the long run, it is not a good thing. The real accumulation of technology is not enough.

Especially with so many projects and clients all at once, the researchers' attention has been greatly divided, and many people are privately complaining about doing things outside of their duties.

The new demands from the government are a good example. Baidu lacks the time to develop the technology, so they have to turn to Meng Fanqi for this requirement.

Wang Kai felt that instead of repeatedly looking for Meng Fanqi, it would be better to just work with him.

Based on his impression of Meng Fanqi's technical prowess, making a lot of money in business is just a matter of how much, and there is no possibility of failure.

After listening to Wang Kai's series of inside information, Meng Fanqi actually understood quite well the thoughts and decisions of the management, which are not very friendly to the lower-level employees.

Because they do not, and cannot really consider the problem from the perspective of the lower-level employees, in other words, they don't care about the thoughts of the workers.

"I don't really have any plans to start a business in the short term, maybe in two years," Meng Fanqi could earn enough wealth just by sharing the technology through technical means, and with some investment, it would be much easier than starting a business.

If he really wants to start his own company, Meng Fanqi currently only has one plan, which is to leave Baidu with Ilya at the end of 2015 and start OpenAI together.

But even this matter has not been thought through, because in the plan, Meng Fanqi would have published the core technologies needed for ChatGPT, such as the attention mechanism and the generative pre-training Transformer, by 2015.

At that time, it might be enough to just invest in OpenAI with technology and capital, be a technical consultant, and there is no need to start a business himself."Let's talk about it later. If I make any moves, I won't hide them. When you see it, remember to contact me."

Someone who had already identified himself so early, Meng Fanqi was also willing to let him join the team, after all, he was also an AI engineer at Baidu, and his level was not a problem.

It's just that he didn't have this need in the short term, and even if he did, it wasn't very polite to directly poach the person sent by Director Li.

After the brief call ended, Meng Fanqi began to search for the Hungarian algorithm and the Kuhn-Munkres (KM) algorithm, both of which were actually old methods.

For example, the Hungarian algorithm was proposed in 1955 and is a combinatorial optimization algorithm that solves allocation problems in polynomial time.

The SORT multi-target tracking method also only uses these two, as well as YOLO proposed by Meng Fanqi, and does not require many changes to these three algorithms themselves.

The YOLO algorithm is used as a feature extractor to extract all parts of the vehicle.

The Hungarian or KM algorithm matches several targets between adjacent frames, and the Kalman filter makes some corrections to the results based on the predicted trajectory of the target.

Especially in government scenes, the movement trajectory of vehicles is relatively regular and single, so the effect of this method will be quite excellent and stable.

In addition, in adjacent frames, Meng Fanqi also introduced the IOU distance, which is the area of intersection between two detection boxes, using it as the weight for bipartite graph matching.

It can also calculate the speed between frames and reasonably design the threshold values of related parameters based on the speed range, which can be said to kill two birds with one stone.

On top of this, there is also an improved version of DeepSORT, which is inspired by the recognition of pedestrians, judging whether two people are the same based on the distance between the parameter vectors of two hidden layers.

However, Meng Fanqi was not prepared to be so thorough. It was enough to meet the basic requirements. He only planned to write down the next idea for Baidu in this part of the continued improvement.

It's more cost-effective to spend time on Baidu.