100 likes | 232 Views
Design and Implement a Face Tracking Engine for Video. Undertaken By: William O’ Grady Project Supervisor and Co Supervisor Dr. Peter Corcoran and Dr. John Breslin. What is Face Detection. A targeted area of object class detection.
E N D
Design and Implement a Face Tracking Engine for Video Undertaken By: William O’ Grady Project Supervisor and Co Supervisor Dr. Peter Corcoran and Dr. John Breslin
What is Face Detection • A targeted area of object class detection. • Object class detection is used daily, such as pedestrian detection in smart traffic lights. • Face detection thus has many applications: • Auto focus for cameras • Security systems • Smart marketing or advertising billboards
Basic Objectives • Research various techniques of face detection on real time video feed • Review most renowned algorithm (Viola - Jones Algorithm) • Design and implement basic algorithm for face detection on real time video feed
Advanced Objectives • Research into improvements in FPS through image pre-processing and face tracking techniques • Implement these techniques and improve on the basic algorithm. • Implement design on embedded device (Android Application) • Develop an evaluation methodology • Produce algorithm which allows real-time tracking at 25+ frame rates or HD video steams (>720p)
Progress to Date • Researched different techniques for face detection • Reviewed current “state-of-art” face detection algorithm • Develop a working algorithm • Tested on live webcam feed • Developed a tracking algorithm for single-frame face detection • Researched into improving algorithm through further pre-processing and/or using multiple images to provide an effective tracking system • Researched various methods for Android implementation
Website • All up to date weekly progress, documents and relevant information can be found here: • http://williamogradyfyp.wordpress.com/
Demonstration • Webcam • Video Feed
Objectives Remaining • Evaluate research of object tracking through pattern matching and/or pre-filtering • Implement design on an embedded device (Android Application) • An evaluation methodology needs to be developed • Develop algorithm which provides real-time tracking at 25+ FPS or on HD video streams (720p or higher) • Develop final design and test on a real-time video feed.
Thanks for your time! • I would be happy to answer any questions that you may have