A multi-granularity parallelism object recognition processor with content-aware fine-grained task scheduling

Abstract

Multiple granularity parallel core architecture is proposed to accelerate object recognition with low area and energy consumption. By adopting task-level optimized cores with different parallelism and complexity, the proposed processor achieves real-time object recognition with 271.4 GOPS peak performance. In addition, content-aware fine-grained task scheduling is proposed to enable low power real-time object recognition on 30fps 720p HD video streams. As a result, the object recognition processor achieves 9.4nJ/pixel energy efficiency and 25.8 GOPS/W·mm2 power-area efficiency in O.13um CMOS technology.