Web-based gaze tracking - OpenCV to JavaScript

Web-based gaze tracking application

We are interested on create a website where the user can watch an image, a video or HTML page and during the experience the system takes the video streaming from the webcam and do the gaze tracking/prediction. In this way, we would be able to collect in a text file the (x,y) position of the gaze in every processed frame.

Note: we saw a prototype developed in C++ using OpenCV on a Linux platform, and we are looking to develop to a web-based application.

We are only interested in a person or a company with objectively verifiable experience in the development of image/video processing routines using web technologies, for example, with experience in integrating OpenCV with Javascript or HTML5 or using other libraries for web-based image processing (JSFeat, Node-OpenCV, etc.).

General users functional requirements:

• PC/notebook webcam, minimum size: 640x480 px.

• Minimum size screen: 14.1 inches (preferably: 13 in.)

• Natural light in the scene, with mainly clear front light.

• Internet browser and version: to be defined or proposed by the developer.

The software must be capable to do (in order of processing):

• Face detection and tracking

• Eyes detection

• Iris detection

• Gaze tracking

The software could ask the user to execute an initial routine for calibrating the system: face location (if the algorithms need the face to be still at the scene), iris location when looking at the center and the four corners.

We are open to suggestions about optional and related web technologies.

We have a detailed project specification, SpreadSheet with desired result records and examples videos of what we are looking for. This material will be provided to the person or company that proves knowledge on the subject.

HTML5 JavaScript

Ταυτότητα Εργασίας: #6664339

Σχετικά με την εργασία