Godkända
Acceleration of Pedestrian Detection System using Hardware-Software Co-design
Mao Hatto (2012)
Start
2013-12-01
Presentation
2014-03-24 00:09
Plats:
E:3139
Avslutat:
2015-03-27
Examensrapport:
Sammanfattning
A demand for human detection is increasing these days, especially for car system or video surveillance. HOG (Histogram of Oriented Gradients) is a core algorithm of such kind of system, and since it is computational intensive algorithm, hardware optimized implementation is demanded. In this work, try to implement human detection system, including hardware optimized HOG algorithm, on FPGA + GPU architecture. Some related research also have implemented on FPGA + GPU system, however, we introduce PEACH2 (PCI-Express Adaptive Communication HUB2) for communication of each node in FPGA + GPU system.
Handledare: Hiderharu Amano (Keito Univeristy, Japan) och Oskar Andersson (EIT)
Examinator: Joachim Rodrigues (EIT)