Fast and Robust Lane Detection using OpenCV
ฝัง
- เผยแพร่เมื่อ 22 พ.ค. 2016
- Using a self-made real time lane detection algorithm, we are able to do fast lane tracking and following. Even if the track is lost for some reason, the car will find it's way back.
The algorithm is written using OpenCV and runs with 30fps on a Intel Atom processor.
If you have any questions, please feel free to contact me! - วิทยาศาสตร์และเทคโนโลยี
@Empty Videos:
It uses the Asus Xtion Pro Live as sensor, but it would also work with a standard webcam.
Processing is done using a Intel Atom processor (board: Kontron pITX E38). The algorithm runs on a single thread, no GPU acceleration. It runs with 30 fps.
Plzzz send me sourcode of this project
@@kratikapatidar8509 what the f....
Brilliant performance, good job
Thanks for sharing. It's really cool
Hi Clemens, what sort of programming language do you use for programming the Intel processor? Also what IDE do you use? This is great! Thanks for sharing
Hi, I'm working on a similar project. Which kind of algorithm did you use for detecting the lanes? I'm trying with binary thresholding and polynomial fitting but it's not that robust.
Did you use convolutional neural network or what?
Seems wonderful. Is there a technical paper stating the theory behind it or is it going to be open sourced?
It is open sourced, just buy a cheap web cam, arduino uno processor, and you'll find the code and guidance in the net.
@@edwinshelly993 u cant use arduino here dude
it dosent have enough computational power lol
@Clemens how can I determine from the Fov of camera the lateral offset and angular offset using camera on the car so that it should not deviate from its lane..
Hi Clemens, what language did you use, and what algorithms ? Also do you have any reference papers that I can use for help, thanks.
Could you describe the Idea of following/tracking the Line or the Path? what type of controller did you use? :)
good work. which microcontroller did you use for the car
Hello sir, can I ask for an overview or tips for this project? thanks
Very cool. Can you share the code and a tutorial on how to set this up. Would be really cool to see the code or for others to try.
Hi Clemens I am trying to make a self driving “car”, I was thinking about using an arduino to run the program, but i’ve been told, that it’s way to slow to manage such calculations. What do you think about it? Could you perhaps help me do it and at least give me some advice, like what hardware did you use (camera or some kind of image recognition, processor unit... and so on). I’m looking forward to your response. Best regards.
It is crossing the line from time to time. Should detect earlier or steer earlier and slow down before the curl
Very nice job.
Agreed.
Agreed
Very good 👏👏
Very nice.
Check this course for an in-depth explanation of OpenCV from beginner to advanced: th-cam.com/video/gLDJEY49M38/w-d-xo.html
What sensors and (processing) board?
Hi I am interested in how you control the car to follow the track. Can you share the project with me? Thank you.
Hello! I would like to know where to find material for building the car ...! And I want to congratulate you on the project, a great job!
Nice are you using hog, haar,.... In open cv? Is your e38 running Linux? is the detection solely based on computer vision?
Super Arbeit! Wie bekommst du diese Verarbeitungszeit hin. Mein Code braucht ganze 200ms um die Linien einer Fahrbahn zu erkennen. Kannst du vielleicht mal beschreiben wie du die Linien suchst? Nutzt du zb auch die Transformation des Bildes in die Vogelperspektive?
Hi! Danke Dir!
Die Linien erkenne ich im perspektivischen Bild und transformiere die Ergebnisse dann in die Vogelperspektive.
Ich mache erst einen dynamischen Schwellwert in jeder Zeile (dass das ganze Kontrastunabhaengig funktioniert) und finde dann Pixel mit hohem Kontrast. Das sind ja Kandidaten fuer Spurenmarkierungen + Rauschen. Durch diese Pixel lege ich mit RANSAC ein Modell (ich denke es war eine Parabel) und transformiere nur die Punkte die in das Modell passen in die Vogelperspektive. Somit verschwindet das meiste Rauschen. Danach vergleiche ich die Punkte mit den Linien die ich in den letzten Bildern erkannt habe und verwerfe Punkte die nicht dazu passen.
Viele Gruesse
at least you share the docs or code of this Wonderful project of yours
Better lane discipline then most human drivers,...
HI is a paper or any code available? Thanks
Sir, I am working on a similar project for my bachelor's degree final project right now, can you share the source code?
Can i buy the car somewhere?
I'm working on something similar. I can detect the lanes but couldn't figure out how to rotate the servo accordingly if the lane has curve the servo should rotate with same angle. ay help? i'm using raspberry pi
did you work it out? I've got same problem too...
read about PID controller
How r u detecting curve lane ..I am able to detect and Mark straight lane but not able to mark curve...plzz suggest
@@gangotrimishra2025 I know its been 2 years but did u find how u can ?
@@soryegetun529 i mean there are several papers and also tutorial named like advance lane detection...you can check those out
Great work! Have you got a paper?
If u find paper plz also send me
do you share the code?
Hello, can you give me a code for reference?
where can I find the program?
Can you please give me your code for reference?
Can I see the code?
Okay, next make it drift around corners😁
Can you make the code available on a github maybe?
The project is great, I want to take a look at the project. Can you give me the code file?
i need this tutorial
So much faster than my old line detection robot... Although to be fair mine was analogue with no programming. Check it out on my oldest TH-cam videos
Nice project do you send me the code source and sensor use
Nice job can you give me more informations about this projet please
Hi Clemens wonderful project could you help me build one with understanding the controllers (hardware) linked to the software .
Do you share source?
Can I get code ??
can you give us the code
How can I contact with you?
I will tell him 12 nights at night, and that Shallah, we will not disagree on the price
If it's 4,5 m/s I'm Cristiano Ronaldo...
4.5 m/s XD lol
HI. Can you send me the code?
Which camera u have used?
Asus Xtion Pro, I wouldn't recommend it anymore though
@@ClemensElflein thank you so much
Sir give me arudino program please 🙏🙏🙏
Great project. may i see ur code?
If not sharing the codes, what is the meaning of showing us this video
Am I missing something? Why is the author not sharing this project? Contribute to learning and development by contributing your work... Is he really that selfish? Hasn't even explained the method or what is used....
Completely fake.
I'll take that as a compliment 😊