JSTS:Journal of Semiconductor Technology and Science

The Institute of Electronics and Information Engineers (대한전자공학회)
권호 표지
DOI QR코드

DOI QR Code

An Energy-Efficient Matching Accelerator Using Matching Prediction for Mobile Object Recognition

  • Choi, Seongrim (Department of Computer Science and Engineering, Chungnam National University) ;
  • Lee, Hwanyong (Department of Computer Science and Engineering, Chungnam National University) ;
  • Nam, Byeong-Gyu (Department of Computer Science and Engineering, Chungnam National University)
  • Received : 2016.03.25
  • Accepted : 2016.03.28
  • Published : 2016.04.30
Download PDF
⟨ Previous Next ⟩

Abstract

An energy-efficient object matching accelerator is proposed for mobile object recognition based on matching prediction scheme. Conventionally, vocabulary tree has been used to save the external memory bandwidth in object matching process but involved massive internal memory transactions to examine each object in a database. In this paper, a novel object matching accelerator is proposed based on matching predictions to reduce unnecessary internal memory transactions by mitigating non-target object examinations, thereby improving the energy-efficiency. Experimental results show a 26% reduction in power-delay product compared to the prior art.

Keywords

References

  1. Y.-C. Su, K.-Y. Huang, T.-W. Chen, Y.-M. Tsai, S.-Y. Chien, and L.-G. Chen, "A 52 mW Full HD 160-Degree Object Viewpoint Recognition SoC With Visual Vocabulary Processor for Wearable Vision Applications," IEEE J. Solid-State Circuits, Vol. 47, No. 4, pp. 797-809, Feb., 2012. https://doi.org/10.1109/JSSC.2012.2185349
  2. J. Joo, S. Choi, J. Ahn, and B.-G. Nam, "Trends in SoC Design for Wearable Devices UI/UX," The Magazine of IEIE, Vol. 41, No. 11, pp. 2-6, Nov., 2014.
  3. D. Nister and H. Stewenius, "Scalable Recognition with a Vocabulary Tree," in Proc. IEEE Computer Society Conf. Computer Vision and Pattern Recogniiton, Vol. 2, pp. 2161-2168, 2006.
  4. Y. Kim, I. Hong, and H.-J. Yoo, "A 0.5V $54{\mu}W$ Ultra-Low-Power Recognition Processor with 93.5% Accuracy Geometric Vocabulary Tree and 47.5% Database Compression," ISSCC Dig. Tech. Papers, pp. 1-3, Feb., 2015.