Csanet High Speed Channel Spatial Attention Network For Mobile Isp

• doi: 10.1109/CVPRW53098.2021.00282
• title: CSANet: High Speed Channel Spatial Attention Network for Mobile ISP
• publisher: IEEE
• isbn: 978-1-6654-4900-7
• issn: 2160-7508
• rank: 1884
• access_type: LOCKED
• content_type: Conferences
• abstract: The Image Signal Processor (ISP) is a customized device to restore RGB images from the pixel signals of CMOS image sensor. In order to realize this function, a series of processing units are leveraged to tackle different artifacts, such as color shifts, signal noise, moire effects, and so on, that are introduced from the photo-capturing devices. However, tuning each processing unit is highly complicated and requires a lot of experience and effort from image experts. In this paper, a novel network architecture, CSANet, with emphases on inference speed and high PSNR is proposed for end-to-end learned ISP task. The proposed CSANet applies a double attention module employing both channel and spatial attentions. Particularly, its spatial attention is simplified to a light-weighted dilated depth-wise convolution and still performs as well as others. As proof of performance, CSANet won 2nd place in the Mobile AI 2021 Learned Smartphone ISP Challenge with 1st place PSNR score.
• article_number: 9523160
• pdf_url: https://ieeexplore.ieee.org/stamp/stamp.jsp?tp=&arnumber=9523160
• html_url: https://ieeexplore.ieee.org/document/9523160/
• abstract_url: https://ieeexplore.ieee.org/document/9523160/
• publication_title: 2021 IEEE/CVF Conference on Computer Vision and Pattern Recognition Workshops (CVPRW)
• conference_location: Nashville, TN, USA
• conference_dates: 19-25 June 2021
• publication_number: 9522011
• is_number: 9522684
• publication_year: 2021
• publication_date: 19-25 June 2021
• start_page: 2486
• end_page: 2493
• citing_paper_count: 12
• citing_patent_count: 0
• download_count: 381
• insert_date: 20210901
• index_terms:
  • ieee_terms:
    • Performance evaluation
    • Runtime
    • Pipelines
    • Network architecture
    • Service-oriented architecture
    • Pattern recognition
    • Image restoration
  • dynamic_index_terms:
    • Spatial Attention
    • Image Signal Processing
    • Processing Unit
    • RGB Images
    • Image Sensor
    • Attention Module
    • Inference Speed
    • Loss Function
    • Convolutional Neural Network
    • Image Quality
    • Mobile Devices
    • Convolutional Layers
    • Feature Maps
    • Part Of Network
    • Attention Mechanism
    • Raw Images
    • Spatial Dependence
    • Residual Block
    • NVIDIA Tesla
    • Spatial Attention Module
    • Channel Attention Module
    • Mobile Camera
    • Channel Attention
    • Conventional Convolution
    • Pixel Loss
    • Running Time Of The Model
    • Perceptual Loss
    • Neural Network
    • Skip Connections
• isbn_formats:
  • format: Print on Demand(PoD) ISBN, value: 978-1-6654-4900-7, isbnType: New-2005
  • format: Electronic ISBN, value: 978-1-6654-4899-4, isbnType: New-2005
• authors:
  • Author Name: Ming-Chun Hsyu
    Affiliation: Industrial Technology Research Institute, Hsinchu, Taiwan, R.O.C
    Author URL: https://ieeexplore.ieee.org/author/37088953313
    ID: 37088953313
    Order: 1
    Author Affiliations:
    • Industrial Technology Research Institute, Hsinchu, Taiwan, R.O.C
  • Author Name: Chih-Wei Liu
    Affiliation: Industrial Technology Research Institute, Hsinchu, Taiwan, R.O.C
    Author URL: https://ieeexplore.ieee.org/author/37089261002
    ID: 37089261002
    Order: 2
    Author Affiliations:
    • Industrial Technology Research Institute, Hsinchu, Taiwan, R.O.C
    • National Yang Ming Chiao Tung University, Hsinchu, Taiwan, R.O.C
  • Author Name: Chao-Hung Chen
    Affiliation: Industrial Technology Research Institute, Hsinchu, Taiwan, R.O.C
    Author URL: https://ieeexplore.ieee.org/author/37088951733
    ID: 37088951733
    Order: 3
    Author Affiliations:
    • Industrial Technology Research Institute, Hsinchu, Taiwan, R.O.C
  • Author Name: Chao-Wei Chen
    Affiliation: Industrial Technology Research Institute, Hsinchu, Taiwan, R.O.C
    Author URL: https://ieeexplore.ieee.org/author/37088951728
    ID: 37088951728
    Order: 4
    Author Affiliations:
    • Industrial Technology Research Institute, Hsinchu, Taiwan, R.O.C
  • Author Name: Wen-Chia Tsai
    Affiliation: Industrial Technology Research Institute, Hsinchu, Taiwan, R.O.C
    Author URL: https://ieeexplore.ieee.org/author/37088952431
    ID: 37088952431
    Order: 5
    Author Affiliations:
    • Industrial Technology Research Institute, Hsinchu, Taiwan, R.O.C

Image Sensor

• sensor_type: CMOS
• resolution: not specified
• 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

Applications & Benefits

• cell_imaging: not specified
• benefits: CMOS image sensors enable digital imaging in smartphones, medical devices, and automotive systems.

Supporting Organizations

• supported_by: not specified

Manuscript Details

• publication_date: 19-25 June 2021

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
  • cell_imaging
  • supported_by
  • publication_date

Processed JSON Filename

request_42380bda-111b-49d0-a88d-f5c22db22f04-csanet_high_speed_channel_spatial_attention_network_for_mobile_isp.json