A 69mW 140-meter/60fps and 60-meter/300fps intelligent vision SoC for versatile automotive applications

Author

Tsai, Yi-Min ; Yang, Tien-Ju ; Tsai, Chih-Chung ; Huang, Keng-Yen ; Chen, Liang-Gee

Author_Institution

DSP/IC Design Lab., Nat. Taiwan Univ., Taipei, Taiwan

fYear

2012

fDate

13-15 June 2012

Firstpage

152

Lastpage

153

Abstract

A machine-learning based intelligent vision SoC implemented on a 9.3 mm² die in a 40nm CMOS process is presented. The architecture realizes 140 meters active distance at 60fps and 60 meters at 300fps under Quad-VGA (1280×960) resolution while maintaining above 90% detection rate for versatile automotive applications. The system supports 64 object tracking and prediction. It raises 1.62× improvement on power efficiency and at least 1.79× increase on frame rate with the proposed knowledge-based tracking processor. The chip achieves 354.2fps/W and 3.01TOPS/W power efficiency with 69mW average power consumption.

Keywords

CMOS digital integrated circuits; automotive electronics; computer vision; electronic engineering computing; learning (artificial intelligence); object tracking; system-on-chip; CMOS process; automotive applications; distance 140 m; distance 60 m; intelligent vision SoC; knowledge-based tracking processor; machine-learning; object tracking; power 69 mW; power consumption; quadVGA resolution; size 40 nm; Calculators; Engines; Feature extraction; Indexes; Knowledge based systems; Shape; System-on-a-chip;

fLanguage

English

Publisher

ieee

Conference_Titel

VLSI Circuits (VLSIC), 2012 Symposium on

Conference_Location

Honolulu, HI

Print_ISBN

978-1-4673-0848-9

Electronic_ISBN

978-1-4673-0845-8

Type

conf

DOI

10.1109/VLSIC.2012.6243835

Filename

6243835