Advanced Driver Assistance System using Detection, Recognition and Tracking of Road Traffic Signs.

This video corresponds to final the results of my research on real time Road Sign Recognition. It was shown on my PhD dissertation 6th July 2009 in the Intelligent Systems Lab. of the Universidad Carlos III de Madrid.
The system is a speed supervisor which broadcasts an audio warning message when the speed of the vehicle is over the limit given by the recognized road sign. When the road sign does not show a speed limit, an individual threshold is used. For example, for yield signs, the speed limit is of 30km/h.
The system can recognize in real time, red and blue road signs both in highways and urban environments under any kind of illumination (not nocturnal). The accuracy of the system is about 0.9.
The system is composed of:
- A camera recording the road scene.
- A computer to run the algorithms for detection, recognition and tracking of the road signs.
- A small screen for the developer which shows the road and the recognitions.
- A PDA which shows the results on recognition and broadcasts the warning messages.
- A GPS inertial sensor to develope the tracking.

The video shows several results:
In the first part the typical problems are shown: false positives and false recognitions, mainly in urban environment which is more problematic than highway environment. The second part shows mainly its performance on highways which is almost perfect.
It is very important to check out that:
- The system recognizes red and blue road signs.
- It works on urban and non-urban environment.
- It works on real time.
- It performs under every day light illumination condition: several images show in the same scene light and darkenss, or back light.
- The system gives the information in advance so the driver can manage an eventual dangerous situation. This may be seen checking both the PDA and the image on board the vehicle.

Category:

Science & Technology

Tags:

• road sign
• traffic sign
• ADAS
• recognition
• computer vision
• detection
• tracking
• Verkehrszeichen
• Erkennung
• Fahrer
• Assistenz System

[jzelar]

hi n1i9c7o5, It is amazing.

I'm a student of San Agustin University from Arequipa-Peru; and I want to ask if you can help me with a university-work, I have to do a similar research, I dont know if you used NN; I suppose you used that. I have to use that for my research and I don`t know exactly the structure from this kind of works; and I want you help me, in that.

[n1i9c7o5]

@jzelar

Hola jzelar, no me queda muy claro donde queires exactamente que te ayude... has leido la informacion que doy sobre el video...? En un sistema como este hay muchas partes que elaborar, te vas a centrar en alguna en concreto? (deteccion, reconocimiento, tracking...)

[gpeetonn]

if it will not disturb you to answer several questions regarding your sign detection system, i will appreciate your help

1. what type of camera did you use for retrieving the image? was it just simple web-camera?

2. what was the resolution of the video? 640x480?

3. was there any noise on image? how did you get rid of it?

4. did you use HSV colour model for thresholding the images of signs on the first step of detection?

5. how did you measure the efficiency of your system?

[mary81yram]

im-presionante....!!!

quien será este tío que se inventa estos aparatejos.....

pero....digo yo, esto está bien para minusválidos o gente mayor al volante....es decir, con que propósito está hecho?? vamos a darle la utilidad social que se merece no?? jejejej

mary