Journals Articles Conferences News About CLP
Submit a paper Email Alert
Search
Search
Articles
Journals
News
Advanced Search
Top Searches
2D MaterialsTransformation opticsQuantum PhotonicsSmall lasersSemiconductor UV PhotonicsSupersymmetric microring laser arrays

Laser & Optoelectronics Progress, Volume. 57, Issue 16, 161005(2020)

Face Recognition Based on Lightweight Neural Network Combining Gradient Features

Xianglou Liu1, Tianhao Li1、*, and Ming Zhang1,2
Author Affiliations
  • 1School of Electronic Science, Northeast Petroleum University, Daqing, Heilongjiang 163318, China
  • 2Heilongjiang University-Enterprise Co-Construction Test and Measurement Technology and Instrument Engineering Research and Development Center, Daqing, Heilongjiang 163318, China
    • show less
    AbstractGet PDF(in Chinese)
    Figures&Tables (9)
    Equations (0)
    References (10)
    Cited By (2)
    Tools
    Paper Information
    Topics
    Figures & Tables(9)
    Flowchart of dimension reduction of Squeeze layer and Expand layer

    Fig. 1. Flowchart of dimension reduction of Squeeze layer and Expand layer

    Download full size
    Flowchart of traditional SqueezeNet model and improved model

    Fig. 2. Flowchart of traditional SqueezeNet model and improved model

    Download full size
    Convergence diagram of training model of the proposed algorithm on LFW dataset for face images with 8×8 blocks

    Fig. 3. Convergence diagram of training model of the proposed algorithm on LFW dataset for face images with 8×8 blocks

    Download full size
    Diagram of ROC

    Fig. 4. Diagram of ROC

    Download full size

    • Table 1. Face recognition rate of the different algorithms for face images with different block sizes%

      View table

      Table 1. Face recognition rate of the different algorithms for face images with different block sizes%

      AlgorithmNo blockBlock 2×2Block 4×4Block 8×8Block 16×16
      SqueezeNet91.6392.1092.1792.9291.19
      SqueezeNet+first-step gradient feature91.9992.6593.1997.2896.97
      SqueezeNet+second-step gradient feature92.6293.4795.7098.3397.39

    • Table 2. Matching time of the different algorithms for face images with different block sizess

      View table

      Table 2. Matching time of the different algorithms for face images with different block sizess

      AlgorithmNo blockBlock 2×2Block 4×4Block 8×8Block 16×16
      SqueezeNet20.9822.1530.2239.2450.86
      SqueezeNet+first-step gradient feature22.1625.4939.7250.3864.73
      SqueezeNet+second-step gradient feature97.02173.15308.83480.973047.31

    • Table 3. Recognition rate of the different algorithms based on each data set with face images of 8×8 blocks%

      View table

      Table 3. Recognition rate of the different algorithms based on each data set with face images of 8×8 blocks%

      AlgorithmLFWCASIA-WebFaceORL
      SqueezeNet92. 9291.9295.49
      SqueezeNet+first-step gradient feature97.2896.5897.27
      SqueezeNet+second-step gradient feature98.3397.7298.96

    • Table 4. Matching time of the different algorithms based on each data set with face images of 8×8 blockss

      View table

      Table 4. Matching time of the different algorithms based on each data set with face images of 8×8 blockss

      AlgorithmLFWCASIA-WebFaceORL
      SqueezeNet39.2423.179.02
      SqueezeNet+first-step gradient feature50.3837.6527.96
      SqueezeNet+second-step gradient feature480.97279.69169.46

    • Table 5. Face recognition rate of cross validation algorithm for face images with different block sizes%

      View table

      Table 5. Face recognition rate of cross validation algorithm for face images with different block sizes%

      AlgorithmNo blockBlock 2×2Block 4×4Block 8×8Block 16×16
      SqueezeNet76.3376.7577.8780.0479.15
      SqueezeNet+first-step gradient feature76.9877.2378.6980.9579.63
    Tools

    Get Citation

    Copy Citation Text

    Xianglou Liu, Tianhao Li, Ming Zhang. Face Recognition Based on Lightweight Neural Network Combining Gradient Features[J]. Laser & Optoelectronics Progress, 2020, 57(16): 161005

    Download Citation

    EndNote(RIS)BibTexPlain Text
    Set citation alerts for article
    Save article for my favorites
    Paper Information

    Category: Image Processing

    Received: Nov. 21, 2019

    Accepted: Jan. 6, 2020

    Published Online: Aug. 5, 2020

    The Author Email: Tianhao Li (2534982997@qq.com)

    DOI:10.3788/LOP57.161005

    Topics

    laser devices and laser physics
    Lasers and Laser Optics
    Laser physics
    laser manufacturing
    Instrumentation, Measurement and Metrology