Cited 0 time in webofscience Cited 0 time in scopus

Real-time Face Tracking in Embedded Systems

Real-time Face Tracking in Embedded Systems
Ahn, Young SunLee, Sang HeonHwang, Byung HunKim, HyundukKim, Yoon Jib
DGIST Authors
Lee, Sang Heon; Kim, Hyunduk
Issue Date
ISIITA 2018, 168-170
In this paper, we propose a fast, accurate face tracking method for real time face recognition in embedded systems. Conventional tracking algorithms, such as Mean Shift, Cam Shift, TLD, and KCF, each has advantages and disadvantages. MeanShift and CamShift are simple operations, but they are only possible if the object and background colors are clearly distinct. TLD and KFC are robust but perform slowly with complex operations. In this paper, we propose a fast, robust face tracker using Template Matching and Haar Cascade to recognize a face in real time in an embedded system. When the face is detected by the face recognizer, the input image and the detected face area are downscaled and the face is found by Haar Cascade within the area surrounding the face position. If a face is not detected using the Haar Cascade, the location of the face is tracked through Template Matching. The experimental results how that, the proposed tracker showed performance similar to that of conventional tracking algorithm, but performed much faster
International Society for Information Technology and Application
Related Researcher
There are no files associated with this item.
Division of Automotive Technology2. Conference Papers

qrcode mendeley

Items in DSpace are protected by copyright, with all rights reserved, unless otherwise indicated.