Face Detection on a parallel platform using CUDA technology

Abstract

Face Detection finds an application in various fields like live video surveillance, bio-metrics, augmented reality,
etc. However, CPU single thread implementation of face detection consumes lot of time, and despite various optimization
techniques, it performs poorly at real time. Our design model combines conventional programming using CPU with multi
core programming using NVIDIA CUDA for fast face detection. The face detection algorithm can be paralyzed to implement
on the GPU using CUDA technology. Viola-Jones face algorithm when exploited in parallel, it gives optimized output. The
speed up factor for various algorithms has been calculated. The algorithms involving rigorous computations can perform
better on parallel platforms when broken up into parallel and executed concurrently.