Neural Network Quantization Is All You Need For Energy Efficient Isp

• doi: 10.1109/EDTM55494.2023.10103005
• title: Neural Network Quantization is All You Need for Energy Efficient ISP
• publisher: IEEE
• isbn: 979-8-3503-3253-7
• issn:
• rank: 3096
• access_type: LOCKED
• content_type: Conferences
• abstract: It is common to devise and develop a Hand Crafted Feature for the modules constituting the CIS ISP Chain. This not only requires specialized domain knowledge, but also has a limitation in the degree of image quality improvement, and there have been many attempts to introduce deep learning to overcome this. However, it is very challenging to circuit the deep learning model directly on the CMOS sensor. Therefore, this study improves energy efficiency through Ultra Low Bit Quantization for demosaicing functions existing in the ISP chain, and proposes an ultra-high-speed/ultra-light model that enables Integer Only Inference. The proposed methodology can be utilized as an element technology that can be applied to deep learning models for other functions constituting ISP.
• article_number: 10103005
• pdf_url: https://ieeexplore.ieee.org/stamp/stamp.jsp?tp=&arnumber=10103005
• html_url: https://ieeexplore.ieee.org/document/10103005/
• abstract_url: https://ieeexplore.ieee.org/document/10103005/
• publication_title: 2023 7th IEEE Electron Devices Technology & Manufacturing Conference (EDTM)
• conference_location: Seoul, Korea, Republic of
• conference_dates: 7-10 March 2023
• publication_number: 10102927
• is_number: 10102928
• publication_year: 2023
• publication_date: 7-10 March 2023
• start_page: 1
• end_page: 3
• citing_paper_count: 0
• citing_patent_count: 0
• download_count: 208
• insert_date: 20230426
• index_terms:
  • ieee_terms:
    • Deep learning
    • Semiconductor device modeling
    • Image quality
    • Quantization (signal)
    • Systematics
    • Neural networks
    • Energy efficiency
  • author_terms:
    • Quantization
    • Demosaicing
    • ISP
  • dynamic_index_terms:
    • Energy Efficiency
    • Neural Network Quantization
    • Deep Learning
    • Deep Learning Models
    • Least Significant Bit
    • Low Bit
    • Layer Model
• isbn_formats:
  • format: Print on Demand(PoD) ISBN, value: 979-8-3503-3253-7, isbnType: New-2005
  • format: Electronic ISBN, value: 979-8-3503-3252-0, isbnType: New-2005
• authors:
  • Author Name: Hyunwoo Je
    Affiliation: SK hynix Inc., Korea
    Author URL: https://ieeexplore.ieee.org/author/37089819937
    ID: 37089819937
    Order: 1
    Author Affiliations:
    • SK hynix Inc., Korea
  • Author Name: Dongil Ryu
    Affiliation: SK hynix Inc., Korea
    Author URL: https://ieeexplore.ieee.org/author/37089818474
    ID: 37089818474
    Order: 2
    Author Affiliations:
    • SK hynix Inc., Korea
  • Author Name: Haechang Lee
    Affiliation: SK hynix Inc., Korea
    Author URL: https://ieeexplore.ieee.org/author/37089821401
    ID: 37089821401
    Order: 3
    Author Affiliations:
    • SK hynix Inc., Korea
  • Author Name: Kijeong Kim
    Affiliation: SK hynix Inc., Korea
    Author URL: https://ieeexplore.ieee.org/author/37089819052
    ID: 37089819052
    Order: 4
    Author Affiliations:
    • SK hynix Inc., Korea

Image Sensor

• sensor_type: CMOS
• resolution: N/A
• dynamic_range: N/A
• pixel_size: N/A
• dark_current: N/A

Optical Data

• focal_length: N/A
• aperture: N/A
• field_of_view: N/A
• distortion: N/A

Performance Metrics

• frame_rate: N/A
• signal_to_noise_ratio: N/A
• sensitivity: N/A
• shutter_speed: N/A
• power_consumption: N/A
• noise: N/A

Applications & Benefits

• cell_imaging: Details about cell imaging applications not specified in the paper.
• benefits: Benefits of using the image sensor not specified in the paper.

Supporting Organizations

• supported_by: SK hynix Inc.

Manuscript Details

• publication_date: 7-10 March 2023

Relevancy Score

• score: 10
• missing_fields:
  • resolution
  • dynamic_range
  • pixel_size
  • dark_current
  • focal_length
  • aperture
  • field_of_view
  • distortion
  • frame_rate
  • signal_to_noise_ratio
  • sensitivity
  • shutter_speed
  • power_consumption
  • noise

Processed JSON Filename

request_025f13ca-61f0-4ee7-9978-22bab940133a-neural_network_quantization_is_all_you_need_for_energy_efficient_isp.json