A 240Fps Incolumn Binarized Neural Network Processing In Cmos Image Sensors

• doi: 10.1109/TCSII.2023.3295391
• title: A 240-FPS In-Column Binarized Neural Network Processing in CMOS Image Sensors
• publisher: IEEE
• isbn:
• issn: 1558-3791
• rank: 369
• access_type: LOCKED
• content_type: Journals
• abstract: This brief presents a CMOS image sensor (CIS) integrated with a binarized neural network (BNN) for face detection in always-on image classification applications. We propose a process variation-immune comparator-based row buffer generating edge images that are inputs of the BNN processor. To reduce the power consumption of column-parallel row buffers, we adopted comparator-based switched capacitor (CBSC) circuits. With a proposed auto-zeroed current source block circuit that operates with low supply voltages, we observed a low variation of row buffers’ outputs. The measurement results showed that the $\sigma /\mu $ of the row buffers’ output is decreased by 4% while reducing 28% of power consumption compared to conventional CBSC-based row buffers. The proposed CIS with an in-column BNN processor having a single channel and two hidden layers was fabricated in a 1-poly 4-metal 110nm CIS process. As a measurement result, we achieved an image classification accuracy is 97.75%. Furthermore, the image resolution is $120\times 120$ , and the total power consumption of the proposed CIS is 3.78 mW with supply voltages of 2.8 V and 1.5 V at 240 frames per second.
• article_number: 10184128
• pdf_url: https://ieeexplore.ieee.org/stamp/stamp.jsp?tp=&arnumber=10184128
• html_url: https://ieeexplore.ieee.org/document/10184128/
• abstract_url: https://ieeexplore.ieee.org/document/10184128/
• publication_title: IEEE Transactions on Circuits and Systems II: Express Briefs
• conference_location:
• conference_dates:
• publication_number: 8920
• is_number: 10262390
• publication_year: 2023
• publication_date: Oct. 2023
• start_page: 3907
• end_page: 3911
• citing_paper_count: 0
• citing_patent_count: 0
• download_count: 642
• insert_date: 20230714
• index_terms:
  • ieee_terms:
    • Image edge detection
    • Power demand
    • Convolution
    • Charge transfer
    • Switching circuits
    • Neural networks
    • Voltage control
  • author_terms:
    • Always-on
    • binarized neural network
    • CMOS image sensor
    • edge mask
    • row buffer
    • face detection
  • dynamic_index_terms:
    • Neural Network
    • Binary Neural Networks
    • Power Consumption
    • Image Classification
    • Current Source
    • Current Regulations
    • Single Channel
    • Supply Voltage
    • Image Edge
    • Face Detection
    • Total Power Consumption
    • Capacitor Circuit
    • Switched Capacitor
    • Low Supply Voltage
    • Convolutional Neural Network
    • Convolutional Layers
    • Feature Maps
    • Sampling Error
    • Analog-to-digital Converter
    • Analog-to-digital
    • Digital-to-analog Converter
    • Analog Voltage
    • Rolling Shutter
    • High Frame Rate
    • Offset Voltage
    • Additional Voltage
• authors:
  • Author Name: Bohyeok Jeong
    Affiliation: Department of Semiconductor Science, Dongguk University (Seoul Campus), Seoul, South Korea
    Author URL: https://ieeexplore.ieee.org/author/37089883549
    ID: 37089883549
    Order: 1
    Author Affiliations:
    • Department of Semiconductor Science, Dongguk University (Seoul Campus), Seoul, South Korea
  • Author Name: Jaehwan Lee
    Affiliation: Department of Industrial and Systems Engineering and the Data Science Laboratory, Dongguk University (Seoul Campus), Seoul, South Korea
    Author URL: https://ieeexplore.ieee.org/author/37089884079
    ID: 37089884079
    Order: 2
    Author Affiliations:
    • Department of Industrial and Systems Engineering and the Data Science Laboratory, Dongguk University (Seoul Campus), Seoul, South Korea
  • Author Name: Suhyeon Lee
    Affiliation: Department of Semiconductor Science, Dongguk University (Seoul Campus), Seoul, South Korea
    Author URL: https://ieeexplore.ieee.org/author/37090012117
    ID: 37090012117
    Order: 3
    Author Affiliations:
    • Department of Semiconductor Science, Dongguk University (Seoul Campus), Seoul, South Korea
  • Author Name: Soyeon Lee
    Affiliation: Department of Semiconductor Science, Dongguk University (Seoul Campus), Seoul, South Korea
    Author URL: https://ieeexplore.ieee.org/author/37090012111
    ID: 37090012111
    Order: 4
    Author Affiliations:
    • Department of Semiconductor Science, Dongguk University (Seoul Campus), Seoul, South Korea
  • Author Name: Youngdoo Son
    Affiliation: Department of Industrial and Systems Engineering and the Data Science Laboratory, Dongguk University (Seoul Campus), Seoul, South Korea
    Author URL: https://ieeexplore.ieee.org/author/37088701320
    ID: 37088701320
    Order: 5
    Author Affiliations:
    • Department of Industrial and Systems Engineering and the Data Science Laboratory, Dongguk University (Seoul Campus), Seoul, South Korea
  • Author Name: Soo Youn Kim
    Affiliation: Department of Semiconductor Science, Dongguk University (Seoul Campus), Seoul, South Korea
    Author URL: https://ieeexplore.ieee.org/author/37859344600
    ID: 37859344600
    Order: 6
    Author Affiliations:
    • Department of Semiconductor Science, Dongguk University (Seoul Campus), Seoul, South Korea

Image Sensor

• sensor_type: CMOS
• resolution: 120x120
• 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: 240 fps
• power_consumption: 3.78 mW

Applications & Benefits

• cell_imaging: Face detection in always-on applications
• benefits: Low power consumption and high frame rate with accurate face detection.

Supporting Organizations

• supported_by: National Research Foundation of Korea (NRF)

Manuscript Details

• publication_date: Oct. 2023

Relevancy Score

• score: 9
• missing_fields:
  • dynamic_range
  • pixel_size
  • dark_current
  • focal_length
  • aperture
  • field_of_view
  • distortion
  • signal_to_noise_ratio
  • sensitivity
  • shutter_speed
  • noise

Processed JSON Filename

request_2fdeb3b7-3624-46a2-a734-5caab1232955-a_240fps_incolumn_binarized_neural_network_processing_in_cmos_image_sensors.json