An Accelerator For High Efficient Vision Processing

• doi: 10.1109/TCAD.2016.2584062
• title: An Accelerator for High Efficient Vision Processing
• publisher: IEEE
• isbn:
• issn: 1937-4151
• rank: 1460
• access_type: LOCKED
• content_type: Journals
• abstract: In recent years, neural network accelerators have been shown to achieve both high energy efficiency and high performance for a broad application scope within the important category of recognition and mining applications. Still, both the energy efficiency and performance of such accelerators remain limited by memory accesses. In this paper, we focus on image applications, arguably the most important category among recognition and mining applications. The neural networks which are state-of-the-art for these applications are convolutional neural networks (CNNs), and they have an important property: weights are shared among many neurons, considerably reducing the neural network memory footprint. This property allows to entirely map a CNN within an SRAM, eliminating all DRAM accesses for weights. By further hoisting this accelerator next to the image sensor, it is possible to eliminate all remaining DRAM accesses, i.e., for inputs and outputs. In this paper, we propose such a CNN accelerator, placed next to a CMOS or CCD sensor. The absence of DRAM accesses combined with a careful exploitation of the specific data access patterns within CNNs allows us to design an accelerator which is highly energy-efficient. We present a single-core implementation down to the layout at 65 nm, with a modest footprint of 5.94mm $^{\boldsymbol {2}}$ and consuming only 336mW, but still about $\boldsymbol {30\times }$ faster than high-end GPUs. For visual processing with higher resolution and frame-rate requirements, we further present a multicore implementation with elevated performance.
• article_number: 7497562
• pdf_url: https://ieeexplore.ieee.org/stamp/stamp.jsp?tp=&arnumber=7497562
• html_url: https://ieeexplore.ieee.org/document/7497562/
• abstract_url: https://ieeexplore.ieee.org/document/7497562/
• publication_title: IEEE Transactions on Computer-Aided Design of Integrated Circuits and Systems
• conference_location:
• conference_dates:
• publication_number: 43
• is_number: 7823113
• publication_year: 2017
• publication_date: Feb. 2017
• start_page: 227
• end_page: 240
• citing_paper_count: 7
• citing_patent_count: 0
• download_count: 1292
• insert_date: 20160622
• index_terms:
  • ieee_terms:
    • Random access memory
    • Biological neural networks
    • Neurons
    • Image recognition
    • Hardware
    • CMOS integrated circuits
    • Computer architecture
  • author_terms:
    • Accelerator architectures
    • convolutional neural network
    • vision sensor
  • dynamic_index_terms:
    • Visual Processing
    • Neural Network
    • Convolutional Neural Network
    • CCD Camera
    • Charge-coupled Device
    • Energy Efficiency
    • Multi-core
    • Quad-core
    • Dual-core
    • CPU Cores
    • Scope Of Application
    • Mining Applications
    • Deep Neural Network
    • Convolutional Layers
    • Feature Maps
    • Neurons In Layer
    • Pooling Layer
    • Single Neuron
    • Processing Elements
    • Current Function
    • Current Map
    • Output Neurons
    • Visual Recognition
    • Classification Layer
    • Input Neurons
    • Output Feature Map
    • Synaptic Weights
    • Read Mode
    • Types Of Layers
    • Motion Estimation
    • Kernel Values
    • Convolutional Neural Network Layers
    • Normalization Layer
    • Average Layer
    • Current Design
    • Current Name
• authors:
  • Author Name: Zidong Du
    Affiliation: University of Chinese Academy of Sciences, Beijing, China
    Author URL: https://ieeexplore.ieee.org/author/37085367291
    ID: 37085367291
    Order: 1
    Author Affiliations:
    • University of Chinese Academy of Sciences, Beijing, China
    • State Key Laboratory of Computer Architecture, Institute of Computing Technology, Chinese Academy of Sciences, Beijing, China
  • Author Name: Shaoli Liu
    Affiliation: State Key Laboratory of Computer Architecture, Institute of Computing Technology, Chinese Academy of Sciences, Beijing, China
    Author URL: https://ieeexplore.ieee.org/author/38066722600
    ID: 38066722600
    Order: 2
    Author Affiliations:
    • State Key Laboratory of Computer Architecture, Institute of Computing Technology, Chinese Academy of Sciences, Beijing, China
  • Author Name: Robert Fasthuber
    Affiliation: EPFL, Lausanne, Switzerland
    Author URL: https://ieeexplore.ieee.org/author/37540520300
    ID: 37540520300
    Order: 3
    Author Affiliations:
    • EPFL, Lausanne, Switzerland
  • Author Name: Tianshi Chen
    Affiliation: State Key Laboratory of Computer Architecture, Institute of Computing Technology, Chinese Academy of Sciences, Beijing, China
    Author URL: https://ieeexplore.ieee.org/author/37402474000
    ID: 37402474000
    Order: 4
    Author Affiliations:
    • State Key Laboratory of Computer Architecture, Institute of Computing Technology, Chinese Academy of Sciences, Beijing, China
  • Author Name: Paolo Ienne
    Affiliation: EPFL, Lausanne, Switzerland
    Author URL: https://ieeexplore.ieee.org/author/37271141000
    ID: 37271141000
    Order: 5
    Author Affiliations:
    • EPFL, Lausanne, Switzerland
  • Author Name: Ling Li
    Affiliation: State Key Laboratory of Computer Architecture, Institute of Computing Technology, Chinese Academy of Sciences, Beijing, China
    Author URL: https://ieeexplore.ieee.org/author/37635092300
    ID: 37635092300
    Order: 6
    Author Affiliations:
    • State Key Laboratory of Computer Architecture, Institute of Computing Technology, Chinese Academy of Sciences, Beijing, China
  • Author Name: Tao Luo
    Affiliation: University of Chinese Academy of Sciences, Beijing, China
    Author URL: https://ieeexplore.ieee.org/author/38548624900
    ID: 38548624900
    Order: 7
    Author Affiliations:
    • University of Chinese Academy of Sciences, Beijing, China
    • State Key Laboratory of Computer Architecture, Institute of Computing Technology, Chinese Academy of Sciences, Beijing, China
  • Author Name: Qi Guo
    Affiliation: State Key Laboratory of Computer Architecture, Institute of Computing Technology, Chinese Academy of Sciences, Beijing, China
    Author URL: https://ieeexplore.ieee.org/author/37306644400
    ID: 37306644400
    Order: 8
    Author Affiliations:
    • State Key Laboratory of Computer Architecture, Institute of Computing Technology, Chinese Academy of Sciences, Beijing, China
  • Author Name: Xiaobing Feng
    Affiliation: State Key Laboratory of Computer Architecture, Institute of Computing Technology, Chinese Academy of Sciences, Beijing, China
    Author URL: https://ieeexplore.ieee.org/author/37632787100
    ID: 37632787100
    Order: 9
    Author Affiliations:
    • State Key Laboratory of Computer Architecture, Institute of Computing Technology, Chinese Academy of Sciences, Beijing, China
  • Author Name: Yunji Chen
    Affiliation: CAS Center for Excellence in Brain Science, Chinese Academy of Sciences, Beijing, China
    Author URL: https://ieeexplore.ieee.org/author/37293328300
    ID: 37293328300
    Order: 10
    Author Affiliations:
    • CAS Center for Excellence in Brain Science, Chinese Academy of Sciences, Beijing, China
    • State Key Laboratory of Computer Architecture, Institute of Computing Technology, Chinese Academy of Sciences, Beijing, China
  • Author Name: Olivier Temam
    Affiliation: INRIA, Palaiseau, France
    Author URL: https://ieeexplore.ieee.org/author/37282483400
    ID: 37282483400
    Order: 11
    Author Affiliations:
    • INRIA, Palaiseau, France

Image Sensor

• sensor_type: CMOS
• resolution: 640x480
• dynamic_range: Not specified
• pixel_size: Not specified
• dark_current: Not specified

Optical Data

• focal_length: Not specified
• aperture: Not specified
• field_of_view: Not specified
• distortion: Not specified

Performance Metrics

• frame_rate: 11 fps
• power_consumption: 336.51 mW

Applications & Benefits

• cell_imaging: Not specified
• benefits: High energy efficiency and performance for visual recognition applications.

Supporting Organizations

• supported_by: NSF of China, 973 Program of China, Strategic Priority Research Program of the CAS, International Collaboration Key Program of the CAS, 10000 talent program.

Manuscript Details

• publication_date: Feb. 2017

Relevancy Score

• score: 8
• missing_fields:
  • dynamic_range
  • pixel_size
  • dark_current
  • focal_length
  • aperture
  • field_of_view
  • distortion
  • signal_to_noise_ratio
  • sensitivity
  • shutter_speed
  • noise
  • microlenses
  • on-chip colour filters
  • global or rolling shutters
  • backside illumination

Processed JSON Filename

request_f541dfa7-3306-444a-89a4-930658fcea10-an_accelerator_for_high_efficient_vision_processing.json