Module for Speaker tracking using OpenCV and Gstreamer

# Speaker tracking using OpenCV and Gstreamer The aim of this project is to create a gstreamer module which will use a Pan, Tilt and Zoom camera to track a person around a stage. The aim is to allow a presentation recording system to operate without needing a person controlling the camera. **When putting in an bid please include links to any previous C or Python code you have produced.** Experience with gstreamer or OpenCV is not required but **highly desirable**. The module will have two modes; * In "detection mode" waving with your hands at the camera should cause the camera to transition into "locked mode" with the person who was waving their hands as the "active person". * In "locked mode" the camera will follow the "active" person keeping their face in the roughly the middle of the screen. If the camera looses the "active" person for over 2 minutes, it should transition back to "detection mode". In locked mode the system should be resistant to the following problems; * Speaker turning away from the camera so that their face is no longer seen. * Speaker moving faster then the camera can follow. * A second speaker who is not the "active" person also being in-front of the camera. * Someone walking in front of the camera causing a **temporary** loss of sight of the speaker. * People's heads appear at the bottom of the camera view. ## Deliverables # Speaker tracking using OpenCV and Gstreamer The aim of this project is to create a gstreamer module which will use a Pan, Tilt and Zoom camera to track a person around a stage. The aim is to allow a presentation recording system to operate without needing a person controlling the camera. **When putting in an bid please include links to any previous C or Python code you have produced.** Experience with gstreamer or OpenCV is not required but **highly desirable**. The module will have two modes; * In "detection mode" waving with your hands at the camera should cause the camera to transition into "locked mode" with the person who was waving their hands as the "active person". * In "locked mode" the camera will follow the "active" person keeping their face in the roughly the middle of the screen. If the camera looses the "active" person for over 2 minutes, it should transition back to "detection mode". In locked mode the system should be resistant to the following problems; * Speaker turning away from the camera so that their face is no longer seen. * Speaker moving faster then the camera can follow. * A second speaker who is not the "active" person also being in-front of the camera. * Someone walking in front of the camera causing a **temporary** loss of sight of the speaker. * People's heads appear at the bottom of the camera view. The module should have the following inputs; * A percentage of view port that the speaker's face should take up. * Maximum speeds for the Pan, Tilt and Zoom changes. * A way to force the module back into "detection mode". The module should have two outputs; * Pan, Tilt and Zoom commands for moving the camera. This output should both be human and machine readable. * The mode which the module is currently in. The code must fulfil the following further requirements: * Be implemented as a gstreamer module. * Run on both Ubuntu Lucid and Ubuntu Precise. * Written in either Python or/and C (not C++). * Should use the OpenCV library. * Should have reasonable level of test cases. * Code developed in this project will be released under the Apache 2 open source license.