The Algorithm method is like bellow. 1 Multi-lane detection is similar to single lane detection, and the algorithm does not limit the number of lanes. 2 Both the straight line and the dotted line are treated as lane lines, and the PCA reduces the type of difference. 3 Steady lane line tracking can be tracked by Kalman, we reverse the PID for error analysis and then track. 4 Without deep learning because it is too slow, the unsupervised learning method is used directly, and the mixed Gaussian model is used specifically, so the overall efficiency increases.
excellent work! i want to learn how can i do this? can you share any paper about this work? what kind of things do u use such as neural networks, opencv, matlab?
Are you finding the lanes directly on 2D image? How you'll find the how much distance lane far away from car? Other than 2D using other stuff such as LiDar or Radar..?
1 Multi-lane detection is similar to single lane detection, and the algorithm does not limit the number of lanes. 2 Both the straight line and the dotted line are treated as lane lines, and the PCA reduces the type of difference. 3 Steady lane line tracking can be tracked by Kalman, we reverse the PID for error analysis and then track. 4 Without deep learning because it is too slow, the unsupervised learning method is used directly, and the mixed Gaussian model is used specifically, so the overall efficiency increases.
Ok :-) this is one of the few multi line trackers i've found and i'm currently doing research on lane positioning given noisy line measurements.. are you aware of others multi-line trackers? bests, augusto
hey bro. your thing is so hard to find... So my team and i decided to make one by ourselves :-) The basic idea is 1 thread for lane detection, 1 thread for lane tracking, and 1 thread for modeling the lane(!!!)
Hello, the lane detection is very stable, I wonder how you do the lane tracking? Do you use any vehicle motion data (such as: IMU, vehicle speed)for lane tracking?
I'ｍ a sensor fusion algorithm engineer from Minieye(a ADAS startup company,this is our web: www.minieye.cc/?lang=en ). We are looking for some talents like you. If you are interested, we can talk about it further. This is my Email: yuji@minieye.cc, waiting for your message.
The Algorithm method is like bellow.
1 Multi-lane detection is similar to single lane detection, and the algorithm does not limit the number of lanes.
2 Both the straight line and the dotted line are treated as lane lines, and the PCA reduces the type of difference.
3 Steady lane line tracking can be tracked by Kalman, we reverse the PID for error analysis and then track.
4 Without deep learning because it is too slow, the unsupervised learning method is used directly, and the mixed Gaussian model is used specifically, so the overall efficiency increases.
excellent work! i want to learn how can i do this? can you share any paper about this work? what kind of things do u use such as neural networks, opencv, matlab?
looks Great! would love to read your paper when it is published. Please notify me when it is published thank a lot.
Impressive - not too different to the visualisations I used to see on my Model 3
Are you finding the lanes directly on 2D image? How you'll find the how much distance lane far away from car? Other than 2D using other stuff such as LiDar or Radar..?
yes, only on 2D image without LiDar or Radar.
Great work, I am a student and we are working on the same thing so please could you upload or give the code, it would be helpful.
Could you please upload the source code
Can u recommend some papers you are refering to for multi-lane detection?
Any paper related to this because this is highly accurate
That's awesome. Waiting for the paper
Sorry bro. I change my research area so the paper was given up by myself. If u want so details u can reach me by mail.
Hi, can i have your email
Hello, great work. Has the paper been published yet? It's two years and a half after...
Hi, Are you done with the Paper writing ? I'm excited to see your paper.
No time for me in recently... Things have to be delay.....
Please how can I train lane detection self driving car ?
Is the paper published?
Would love to read it!
Can you share the code? Or any reference?
Hi, Mr Chan, I am working on lane line detection, too. May I ask for your unpublished paper as a detailed reference?
1 Multi-lane detection is similar to single lane detection, and the algorithm does not limit the number of lanes.
2 Both the straight line and the dotted line are treated as lane lines, and the PCA reduces the type of difference.
3 Steady lane line tracking can be tracked by Kalman, we reverse the PID for error analysis and then track.
4 Without deep learning because it is too slow, the unsupervised learning method is used directly, and the mixed Gaussian model is used specifically, so the overall efficiency increases.
Great work! Where can i find the link to paper ?
Hello, Mr.Chan, did you finish your paper, post a link here if you did, thanks !
Sorry bro. I change my research area so the paper was given up by myself. If u want so details u can reach me by mail.
Very nice one, Mr.Chan, could you pls send me some more details about your algorithm to zhouwu005@163.com, by the way which area did you transfer to?
Sanjay chan hi,can you please send me some more details about your algorithm to me by email tielang2253@hotmail.com?thanks!
hi mr chan can u email me your research paper at me thariq1108@gmail.com
malikmanan422@gmail.com
Hello Mr. Sanjay chan
, I'm new to this research area.
I know It's difficult but could you give me access to your code please?
my email: ryan.vt1997@gmail.com
Do you use deep learning method?? such as semantic segmentation based deep learning??
sorry for your curiosity. in order to overcome the overfitting problem, this algorithm is base on Unsupervised learning.
could you give some hints which Unsupervised learning method in your lane detection and tracking algorithms?
1多车道检测跟单车道检测类似，算法不限制车道数量。
2直线和虚线都当做是车道线，后面用PCA降维区别类型。
3稳定车道线跟踪可以用卡尔曼跟踪，我们逆用PID进行误差分析然后跟踪。
4没有用深度学习因为太慢，这里直接用无监督的学习方法，具体使用高斯模型，所以整体效率就上升了。
Hi Sanjay chan,
any news about the paper or the source code?
and which library did u use? OpenCV I guess
how you ignore the arrows on the middle in lane detection? do you depend on Hough transform to find lines?
perfect, base on C++ and OpenCV.
@@ChannelCtrlAltDefeatso wrong....
I have noticed the chinese characters in the data, so may I ask what dataset do you use?
we collect by ourselves.....
Hi, great work! do you plan to freely release an implementation of your algorithm?
Maybe after I leave current company...
Ok :-) this is one of the few multi line trackers i've found and i'm currently doing research on lane positioning given noisy line measurements.. are you aware of others multi-line trackers? bests, augusto
hey bro. your thing is so hard to find... So my team and i decided to make one by ourselves :-) The basic idea is 1 thread for lane detection, 1 thread for lane tracking, and 1 thread for modeling the lane(!!!)
@@ZOMI666 did you leave your company yet??? I would love to have a look at your code.
it is great. Is it deep learning based method？
It is base on Unsupervised learning
Hi, may I ask if this algorithm uses machine learning approach?
absolutely
Is it of graphical model or deep learning?
Is it of graphical model or deep learning?
hey bro. the algorithm is :-) 1 thread for lane detection, 1 thread for lane tracking, and 1 thread for modeling the lane.
can you share me your code? tkank you very much
Did you finally manage to write the associated paper ?
Sorry bro. I change my research area so the paper was given up by myself. If u want so details u can reach me by mail.
hi. i am very interesting this research. so i wanna read some paper. would you recommend it?
Is there Taiwan?
it's in mainland China
Please 🥺 shall I have the code
student or professor of south China agricultural university?
sorry, i already graduated nearly 3 years.
Sanjay chan 您好，是这样子的，最近我在做一个跟路线检测有关的项目，但是却无从入手。请问您可以为我知道一下方向吗？
Sanjay chan 我也是华农的学生，师兄你好😂😂😂
@@wongandy3239 maybe ,South China University of Technology........
please send code
大佬，开源吗
did you try it at night
I Have try, and the result is great.
@@ZOMI666 can you share the source code with me sir sohardef@gmail.com
Hello, the lane detection is very stable, I wonder how you do the lane tracking? Do you use any vehicle motion data (such as: IMU, vehicle speed)for lane tracking?
I want to use those stuff. but nothing in the algorithm.
Great job. Do you test the algorithm on the quick lane change data?
of course. but your question were quick a bit ambiguous. Point is depend how quick you mean....
I'ｍ a sensor fusion algorithm engineer from Minieye(a ADAS startup company,this is our web: www.minieye.cc/?lang=en ). We are looking for some talents like you.
If you are interested, we can talk about it further. This is my Email: yuji@minieye.cc, waiting for your message.
Can u send me video clip?
you can download it though www.clipconverter.cc/
of course. leave your mail.
Hi, can I have the original video too? ma.huifang0604@gmail.com
hello two guys. there are too many video clip like these in my dataset. every one is so big in GB. so I am not sure which one you like.
I want original video. So I can test my lane detection
Can you give me the details ? As I get to know that you gave up on the research papers
check out Sentdex Python plays series... this is prob cv2
Nice one
这个给力啊
放源码啊
など私は論文を完成し終わって