Traffic lights identify open datasets traffic Lights recognition (TLR) public benchmarks

Traffic Lights Recognition (TLR) public benchmarks Urban Scene 1 Dataset

©read the copyrights information before any use.

On-board vehicle acquisition in a dense urban environment.

179 Frames (8min 49sec, @25fps)
640x480 (RGB, 8bits)
Paris (France)

Acquisition Description:
Acquired from the C3 vehicle, Camera sensor marling f-046c (@ 25Hz), lens 12mm, camera is mounted behind the interior rea R-view Mirror, vehicle speed is < 50km/h (< mph)


Downloads

Sequence and groundtruths data is available publicly and for free.
The sequences can downloaded either as MPEG-2, JPEG single files, Jseq, or rtmaps (cf. below). Ground truths can also is downloaded from the same page. Since it exists several file format for Ground Truth we chose to distribute we have files in all the main formats:gt (text fo rmatting), CVML, and VIPER.

Data is also available as rtmaps files which contain raw acquisition data (such As:camera output with timestamps data). Rtmaps is a real time multisensor prototyping software which we use as on-board application to record our acquisitions and Then to replay the latter. More information is available on the Rtmaps Compagny website.

We'll be pleased to publish the result of your traffic light recognition algorithmin our website. As long as you use the same databases (or if your databases is public).

Sequence 11179 frames (640x480, RGB, 8bits) Download JPEG single files (Zip, 468MB) Download MPEG2 sequence (mpg, 208MB) Download jseq sequence (zip, 478MB) Download REC rtmaps Acquisition data (Zip, 478MB)

Ground Truth files v0.5 (9168 hand-labeled traffic lights) Download groundtruth GT Format (TXT, 0.5MB) text format, data separated by Space Download groundtruth cvml Format (XML, 4 MB) XML format, more information Download groundtruth VIPER Format (XML, 0.5MB) XML format, more infor Mation Benchmarks

Here is listed the performance of the algorithms on the above described sequences. For more information on the evaluation, please refer to FAQ section below.
If you want your algorithm to being listed in this section contact us and send us your result (cf. publishing your result s). Robotics Centre of Mines ParisTech and Imara Team of Inria (May, 1st)

(Raoul de Charette1 and Fawzi Nashashibi1, 2, 2010)

Download High res. Video 1min44 (XVID, 240MB)
Download low res. Video 1min44 (XVID, 20MB)

Publications
[1] R. de Charette and F. Nashashibi, "Real Time visual traffic lights recognition b ased on Spot light Detection and adaptive traffic lights templates, " 2009 IEEE Intelligent Vehicles Symposium, Xian: IEEE, P, pp. 358-363. (Read ON&NBSP;IEEEXPLORE&NBSP;-&NBSP;ACM)
[2] R. de Charette and F. Nashashibi, "traffic light recognition using image processing compared to learning Processe S, " 2009 IEEE/RSJ International Conference on Intelligent Robots and Systems (IROS), Saint Louis:ieee,, pp. 333 -338. (Read ON&NBSP;IEEEXPLORE&NBSP;-&NBSP;ACM)
Please note these publications does not describe the current state of our traffic light recognition system. It has evolved consequently since these, publications.
A New publication would be published describing the whole system.

1 Robotics Centre of Mines ParisTech, France (Caor-centre de Robotique)
2 Imara Team, Inria Rocquencourt, France (Imara-informatique, mathématiques et automatique pour la Route automatisée) Electrical and computer Engineering Department, University of Patras, Rio, Patras, Greece (+)

[1] G. Siogkas, E. Skodras, and E. Dermatas, "Traffic Lights Detection in adverse Conditions Using Color, symmetry and Spa Tiotemporal information, "in International Conference on computer Vision theory and Applications (Visapp), pp. 620–627. (Read on Patra ' s website-research Gate)
FAQ How can I add the performance of my algorithm on this page?

Please refer the sections publishing your results. which is the objects labeled in the sequence?

so far, only traffic Lights (with circular light). But since we made this sequence publicly, if you is a do-gooder feel free to label others objects in the sequence and to SE nd us the new ground truth file that we'll be pleased to add in this webpage. How many objects is labeled in the Ground Truth file?

The ground Truth file contains 9 168 instances of traffic lights, hand-labeled.
Traffic lights detail is as follows: 3 381 "green" (called ' Go '), "Orange" (called ' Warning '), 5 280 "Red" (called ' Stop '), 449 "ambiguous" (cf. below). What is called a "ambiguous traffic light"?

During The labeling process our human operator noticed several ambiguous regions for which they had issue to decide whethe R It is a real traffic light (with circle light) or not. We thus decided to simplyignored these ambiguous regionsDuring the evaluation. Therefore, any traffic light detected in these regions won ' t being taken into account neither as a "false positive" nor as a "True positive".
Indeed, there is very few number of "ambiguous" regions and they were strictly labeled "ambiguous" only if they validate One of the following conditions:
Reflection Distortion. The region was a reflection of an object which seems to being a traffic lightLight shape not valid. The light of the traffic light appears circle were it's in fact a rectangle (usually due to CCD approximation or motion B Lur)Too Blurry. The traffic light is ' too ' blurry during it whole timeline (usually due to vehicle turning, vehicle pitch, or potholes) ( For instance, frames 3568-3616)Too Small. The traffic light was too small during its whole timeline. (For instance, frame 9 200)Not facing the vehicle. The traffic light isn't facing the vehicle but the light is still visible. (For instance, frame 9 302)Lower Traffic Light. The small and lower traffic lights under the big one is ignored. The latter is specific to France (for instance, frame 5 260)What is the minimum size of labeled traffic lights?

Traffic lights were labeled as soon as they is 5 pixels wide ormore. Why is there objects with coordinates off-limit?

Coordinates out-of-bounds (negative or superior than image width/height) is due to the  traffic lights PA rtially Visible   (leaving the camera Field of View). These occluded traffic lights is ignored during the evaluation.