Title :
Real-Time Object Recognition with Neuro-Fuzzy Controlled Workload-Aware Task Pipelining
Author :
Kim, Joo-Young ; Kim, Minsu ; Lee, Seungjin ; Oh, Jinwook ; Oh, Sejong ; Yoo, Hoi-Jun
Author_Institution :
Korea Adv. Inst. of Sci. & Technol., Daejeon, South Korea
Abstract :
A proposed object recognition processor lightens its workload by estimating global region-of-interest features. A neuro-fuzzy controller performs intelligent ROI estimation by mimicking the human visual system, then manages the processor´s overall pipeline stages using workload-aware task scheduling and applied database size control. The NFC performs workload-aware dynamic power management to reduce the proposed processor´s power consumption.
Keywords :
fuzzy control; neurocontrollers; object recognition; pipeline processing; global region-of-interest feature; human visual system; intelligent ROI estimation; neuro-fuzzy control; object recognition processor; real-time object recognition; workload-aware dynamic power management; workload-aware task pipelining; workload-aware task scheduling; Control systems; Deductive databases; Humans; Object recognition; Pipeline processing; Power system management; Processor scheduling; Size control; Spatial databases; Visual system; multicore processor; neuro-fuzzy controller; object recognition; three-stage task pipelined architecture; visual perception; workload-aware dynamic power management;
Journal_Title :
Micro, IEEE
DOI :
10.1109/MM.2009.102
