Yang Chaoxiong didn't have any objections to Senior Brother calling Ye Ming over. His mind was now full of bugs in the current test card.
"Have you read the project report?"
"Not yet, I just went to read the paper."
"Well, the topic of our group is the machine vision solution in the assisted driving system. Chaoxiong is responsible for the dynamic perception of the environment by the binocular camera."
While introducing, Ma Jun paid attention to Ye Ming's expression. When he saw that the latter just nodded slightly and showed no timidity, he smiled slightly.
It seems that the teacher did not arrest the wrong person. This guy at least has confidence.
"Stop talking, come and help me take a look." Yang Chaoxiong stood up and gave up his seat to Ma Jun.
Ma Jun sat down and dragged the mouse a few times. He only looked at the report and then turned around and smiled: "You guy, you wrote the algorithm, you built the framework and environment, I can see that you are confused.
?”
Yang Chaoxiong looked helpless: "...Brother, I really checked it for a long time, and I just had to use a magnifying glass to check the source code."
"If you were a teacher, he would definitely ask you to take a magnifying glass and look at the source code."
"Brother, there are tens of thousands of lines of code."
"Then you have to masturbate." Ma Jun chuckled: "Is your algorithm okay?"
"If there is a problem with the algorithm, it won't run, brother." Yang Chaoxiong was obviously unwilling to show the source code, let alone admit that there was a problem with his algorithm. He hesitated and said: "How about changing to a simulation engine? I suspect that the engine is working against me."
"..."
…
Ye Ming was listening to the conversation between the two, and with Ma Jun explaining on the side, he understood what Yang Chaoxiong's mission was.
In the current machine vision perception test of autonomous driving, there are generally three methods. One is based on the software engineering method, building a model to input data for simulation testing. The other is using the virtual simulation method, using the simulation engine to generate three-dimensional game screens.
Construct a virtual scene to test the algorithm.
The last method is to directly put the machine vision perception system in the car and drive to the real site for testing.
Each of these three methods has its own advantages and disadvantages. Although the third method is the closest to reality and has reliable data, its shortcomings are obvious, that is, it cannot meet the diversity of scenarios.
Therefore, most people engaged in autonomous driving use the second method to conduct testing in order to obtain relatively accurate data and diversity of scenarios.
Yang Chaoxiong also used the second method - the graphics card of his computer is a 3090Ti.
The current problem is that when Yang Chaoxiong runs the test, he always cannot pass the trajectory recognition of objects in relative motion - to put it simply, after Yang Chaoxiong's algorithm starts running, someone throws a stone, or the car in front drops a cardboard
Something like that, the algorithm can't determine the landing point.
And it’s not that I can’t tell it every time, it’s just occasionally.
This is a little crazy.
Hmm... Although Ye Ming doesn't know where the problem lies.
But he knows how to solve the problem.
Rewrite one.
…
This is when Ma Jun suddenly looked at Ye Ming: "Ye Ming, do you understand the project that senior brother is responsible for?"
Ye Ming was slightly startled.
Aren’t you here to make soy sauce?
Do you have any other matters of your own?
"I probably understand."
"What do you think?" Ma Jun looked at him with a smile and said meaningfully: "The teacher said that you are very talented in machine vision."
Ma Jun's words made Yang Chaoxiong on the side suddenly startled and looked at Ye Ming fiercely.
Ye Ming... was actually surprised.
But he immediately shook his head and said with a smile: "I have no idea. Compared with Senior Brother Yang's environmental perception, the machine vision I did is just like a child playing house."
——What he said is indeed correct.
Although he has been dealing with the trajectories of moving objects for a full month including the holidays, his methods are so simple... Whether it is a table tennis ball, a bullet, or an RC ball, parabola calculations
You'll know it right away. At most, you can add speed recognition. Anyway, it all follows the laws of physics.
But here we are engaged in autonomous driving environment perception. When you are driving on the road, who knows what will fly over?
Cardboard, stones, bricks, tires, even towels...
There will always be strange things you don't expect flying towards your face.
What autonomous driving has to do is to identify these things and determine whether these things will fly to the face, so as to decide what action the car will take.
The level of difficulty is unparalleled.
After hearing this, Yang Chaoxiong's expression relaxed.
This is right.
No matter how you do self-aiming or adjust parameters, it is impossible to compare with the environment perception of autonomous driving!
Ma Jun also chuckled: "Well, how about you team up with Chaoxiong - I asked the teacher at the beginning, and the teacher also suggested the same."
Ye Ming looked at Yang Chaoxiong, and saw that the latter was also looking at him. The two looked at each other, and seemed to feel that the other party liked each other. They smiled in understanding and nodded at the same time.
In the next few days, Ye Ming began to go to the competition preparation hall at night to prepare the plan, and during the day to go to the research team to read literature, improve the code, and write the paper.
Of course, he will also follow Yang Chaoxiong for a while, playing tricks, running code and so on.
Then, in the second week, he had his first research group meeting in his life.
…
Sitting in the last seat, Ye Ming was almost facing the serious face of Professor Tang Zhigao, which made him immediately realize why when Ma Jun talked about holding a group meeting, these senior brothers and sisters who were usually laughing and joking would
It changed color instantly.
It seems that the rumors about Lao Tang’s rigorous scholarship are really not a legend.
At the group meeting, Professor Tang commented on each person's report one by one. Although he did not curse, it was obvious that he was not very satisfied with the progress of the current project.
After the graduate students finished reporting one by one, Professor Tang looked at Ye Ming.
Then everyone saw that the expression on Professor Tang's face, who had been so majestic just now, suddenly softened.
"Ye Ming, what about you?"
"Teacher Tang, I didn't prepare PPT."
"You have no scientific research tasks, so which PPT should you use?" Professor Tang Zhigao motioned Ye Ming to sit down directly: "Just talk about your experience these days."
"Well, I have been reading literature and learning machine learning and machine vision from Brother Yang these days."
Tang Zhigao nodded slightly.
After pondering for a moment, he smiled and asked, "How did you learn?"
"Uh..." Ye Ming was speechless for a moment.
How should he answer?
I can actually know what Senior Brother Yang knows?
Even if I don’t know how to do it, can I still make a trick and force me to know how to do it?
Just when he was about to make a fuss and show that he was working hard, Professor Tang suddenly smiled.
"Chaoxiong will be on a business trip with me for the next half month. You have studied with him for so many days, and you should be almost familiar with him. In the next half month, you will be responsible for taking over Chaoxiong's tasks."
Everyone's eyes suddenly stopped on Ye Ming.
Will first-year undergraduates take over the tasks of first-year students?