A Near-Sensor Processing Accelerator for Approximate Local Binary Pattern Networks

Shaahin Angizi, Mehrdad Morsali, Sepehr Tabrizchi, Arman Roohi

Research output: Contribution to journalArticlepeer-review

1 Scopus citations


In this work, a high-speed and energy-efficient comparator-based <underline>N</underline>ear-<underline>S</underline>ensor <underline>L</underline>ocal <underline>B</underline>inary <underline>P</underline>attern accelerator architecture (NS-LBP) is proposed to execute a novel local binary pattern deep neural network. First, inspired by recent LBP networks, we design an approximate, hardware-oriented, and multiply-accumulate (MAC)-free network named Ap-LBP for efficient feature extraction, further reducing the computation complexity. Then, we develop NS-LBP as a processing-in-SRAM unit and a parallel in-memory LBP algorithm to process images near the sensor in a cache, remarkably reducing the power consumption of data transmission to an off-chip processor. Our circuit-to-application co-simulation results on MNIST and SVHN datasets demonstrate minor accuracy degradation compared to baseline CNN and LBP-network models, while NS-LBP achieves 1.25 GHz and an energy-efficiency of 37.4 TOPS/W. NS-LBP reduces energy consumption by 2.2&#x00D7; and execution time by a factor of 4&#x00D7; compared to the best recent LBP-based networks.

Original languageEnglish (US)
Pages (from-to)1-11
Number of pages11
JournalIEEE Transactions on Emerging Topics in Computing
StateAccepted/In press - 2023

All Science Journal Classification (ASJC) codes

  • Computer Science (miscellaneous)
  • Information Systems
  • Human-Computer Interaction
  • Computer Science Applications


  • Processing-in-memory
  • SRAM
  • accelerator
  • near-sensor processing


Dive into the research topics of 'A Near-Sensor Processing Accelerator for Approximate Local Binary Pattern Networks'. Together they form a unique fingerprint.

Cite this