Visual attention & Eye-tracking
This platform allows to work with material and software which record the eye gaze and analyze it to generate saliency map or other information on visual attention. All of our hardware can be used to generate saliency maps for each observer and/or for all of them. This truth can be compared to saliency map generated by automatic visual attention models or used for some other studies. The data of both eyes can be analyzed to measure the effect of 3D content visualisation for exemple. Another possibility is to evaluate the path of the strong eye compare to the path of the other eye. All these ressources allow to measure the impact of the degradation position, the area of interest to have a robust protection or whatever.
It can be used for images, videos or 3D contents to create saliency map or to view the differences between the both eyes. It is the most accurate of our eye-tracking systems (between 0.25° and 0.5°). It can work in binocular mode up to 500Hz. It aslo can go up to 1250Hz in monocular mode to have a very good temporal precision when the goal is to measure some psycho-visual properties. The materiel can be used with the SMI software or with a custom software and the SDK.
The remote eye-tracking system allow to track eyes without materiel on the head. Little movments of the head can be done but the observer must stay on his chair and not move a lot because the system is very sensitive. It has a good accuracy (0.4°) and works on binocular eyes with 60Hz frequency.
The SMI ETG is an eye-tracking glasses which allow movments and so, can be used more friendly with the natural movments of the head for exemple. It is coupled with ARENA, a headtracker solution which can capture the position of markers with infrared motions cameras and so, gives a gaze vector in the world coordinate system of ARENA. Finally, in real time, our custom software, with a calibration function, manage to get the gaze point in the screen coordinate system which is what we needed.
The system is a binocular eye-tracker with a frequency of 30Hz. Our studies demonstrate that it is a little bit less accurate that our others eye-trackers in the screen center region but has the same accuracy in the border regions of the screens (and the other eye-trackers are very poor in this region). So it's works very well for large screen (for exemple, for UHD screens) and the other advantage is the fact that the observer can be move more naturally than our other eye-trackers.
We also have one other head-mounted system and the Cambridge eye-tracker which works in monocular mode with 50Hz. All of the systems are synchronized with the displaying of images, videos from QCIF (or less) to HD and 3D contents (images and videos).
Eyetracker_SD_2009_12 Eye tracking and quality assessment (with the ACR-HR method) on SD videos
This database is composed by 20 original SD videos and 80 distorted videos by H264 coding with transmission errors. The H264 streams, subjective scores (ACR-HR) and eyetracker raw data of 30 observers are available. Read more.
HD_UHD_Eyetracking_Videos 37 videos viewed in HD or in UHD with eyetracking data
A new HD and UHD video eye tracking dataset composed of 37 high quality videos observed by more than 35 naive observers. This dataset can be used to compare viewing behavior and visual saliency in HD and UHD, as well as for any study on dynamic visual attention in videos. Read more.
Intentionally_Blurred_Background_Images Eye-tracking saliency maps and importance maps for blurred images
This image database contains 72 images. There are 6 versions of each content : the original and 5 power of blur. The database also contains eye-tracking data of 18 observers for these images. Read more.