An Onchip Binaryweight Convolution Cmos Image Sensor For Neural Networks

• doi: 10.1109/TIE.2020.3001838
• title: An On-Chip Binary-Weight Convolution CMOS Image Sensor for Neural Networks
• publisher: IEEE
• isbn:
• issn: 1557-9948
• rank: 1519
• access_type: LOCKED
• content_type: Journals
• abstract: A CMOS image sensor (CIS) that can perform on-chip binary convolution is presented. The CIS can greatly reduce memory usage and computational complexity by directly generating a feature map for a binary neural network. The pixel readout of the CIS is performed in the column-parallel fashion using incremental delta-sigma analog-to-digital converters (ADCs). The CIS operates in two different modes: convolution and normal modes. When the column ADC is working in the convolution mode, it works as a first-order delta-sigma ADC and generates convolved images using a binary kernel. In the normal operation mode, the ADC is switched to a second-order delta-sigma ADC with little hardware modification and used to capture high-quality images. To demonstrate the CIS architecture, a 192 × 128-pixel CIS, which occupies an active die area of 14.44 mm2, is fabricated in a 0.18 μm standard CMOS process. The performance of the CIS is evaluated through measurements and network simulations. In the normal operation mode, the CIS achieves a read noise of 14.79 e-rms and a full-well capacity of 6,420 e- with a resulting dynamic range of 53 dB. The power consumptions of the CIS are 49.2 and 52.5 mW during the normal and convolution modes, respectively.
• article_number: 9119774
• pdf_url: https://ieeexplore.ieee.org/stamp/stamp.jsp?tp=&arnumber=9119774
• html_url: https://ieeexplore.ieee.org/document/9119774/
• abstract_url: https://ieeexplore.ieee.org/document/9119774/
• publication_title: IEEE Transactions on Industrial Electronics
• conference_location:
• conference_dates:
• publication_number: 41
• is_number: 9420632
• publication_year: 2021
• publication_date: Aug. 2021
• start_page: 7567
• end_page: 7576
• citing_paper_count: 12
• citing_patent_count: 0
• download_count: 1611
• insert_date: 20200617
• index_terms:
  • ieee_terms:
    • Kernel
    • Convolution
    • Quantization (signal)
    • Modulation
    • Neural networks
    • Field programmable gate arrays
    • Application specific integrated circuits
  • author_terms:
    • Binary weight neural network
    • CMOS image sensor
    • convolution operation
    • incremental delta-sigma analog-to-digital converter
  • dynamic_index_terms:
    • Neural Network
    • Feature Maps
    • Operation Mode
    • Analog-to-digital Converter
    • Analog-to-digital
    • Digital-to-analog Converter
    • Memory Usage
    • Simulated Networks
    • Hardware Modifications
    • Convolutional Network
    • Convolutional Neural Network
    • Network Layer
    • Convolution Operation
    • Convolution Kernel
    • Classification Rate
    • Classification Evaluation
    • Digital Signal
    • Root Mean Square Values
    • Rms Value
    • Root-mean-square Values
    • Quantization Error
    • Amount Of Error
    • Quantization Noise
    • Circuit Implementation
    • Output Pixel
    • Digital Output
    • Kernel Values
    • Readout Circuit
    • Integrator Output
    • Tungsten Halogen
    • Tungsten Halogen Lamp
    • Tungsten-halogen
    • Tungsten-halogen Lamp
    • Tungsten Halogen Light
    • Digital Integration
    • Parasitic Capacitance
    • Input Samples
    • Noise Amplification
• authors:
  • Author Name: Woo-Tae Kim
    Affiliation: School of Electrical Engineering and Computer Science, Gwangju Institute of Science and Technology, Gwangju, South Korea
    Author URL: https://ieeexplore.ieee.org/author/37086270747
    ID: 37086270747
    Order: 1
    Author Affiliations:
    • School of Electrical Engineering and Computer Science, Gwangju Institute of Science and Technology, Gwangju, South Korea
  • Author Name: Hyunkeun Lee
    Affiliation: School of Electrical Engineering and Computer Science, Gwangju Institute of Science and Technology, Gwangju, South Korea
    Author URL: https://ieeexplore.ieee.org/author/37086263856
    ID: 37086263856
    Order: 2
    Author Affiliations:
    • School of Electrical Engineering and Computer Science, Gwangju Institute of Science and Technology, Gwangju, South Korea
  • Author Name: Jung-Gyun Kim
    Affiliation: School of Electrical Engineering and Computer Science, Gwangju Institute of Science and Technology, Gwangju, South Korea
    Author URL: https://ieeexplore.ieee.org/author/37086919924
    ID: 37086919924
    Order: 3
    Author Affiliations:
    • School of Electrical Engineering and Computer Science, Gwangju Institute of Science and Technology, Gwangju, South Korea
  • Author Name: Byung-Geun Lee
    Affiliation: School of Electrical Engineering and Computer Science, Gwangju Institute of Science and Technology, Gwangju, South Korea
    Author URL: https://ieeexplore.ieee.org/author/38240871700
    ID: 38240871700
    Order: 4
    Author Affiliations:
    • School of Electrical Engineering and Computer Science, Gwangju Institute of Science and Technology, Gwangju, South Korea

Image Sensor

• sensor_type: CMOS
• resolution: 192x128 pixels
• dynamic_range: 53 dB
• pixel_size: Not specified
• dark_current: Not specified

Optical Data

• focal_length: 8 mm
• aperture: Not specified
• field_of_view: Not specified
• distortion: Not specified

Performance Metrics

• power_consumption: 49.2 mW (normal mode), 52.5 mW (convolution mode)
• noise: 14.79 e-rms

Applications & Benefits

• cell_imaging: Used in image recognition applications such as self-driving vehicles, surveillance, biomedical diagnosis.
• benefits: Reduces memory usage and computational complexity by generating feature maps directly on the chip.

Supporting Organizations

• supported_by: MOTIE Research Grant 2020, MSIT (Ministry of Science and ICT), Korea

Manuscript Details

• publication_date: Aug. 2021

Relevancy Score

• score: 10
• missing_fields:
  • pixel_size
  • fill_factor
  • quantum_efficiency
  • dark_current
  • readout_speed
  • noise_sources
  • microlenses
  • on-chip colour filters
  • global or rolling shutters
  • backside illumination

Processed JSON Filename

request_1f361dfc-648b-45b8-9361-2c636b931711-an_onchip_binaryweight_convolution_cmos_image_sensor_for_neural_networks.json