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

Acta Optica Sinica, Volume. 39, Issue 2, 0210002(2019)

Multi-View Indoor Human Detection Neural Network Based on Joint Learning

Xia Wang1,2、* and Wei Zhang1,2
Author Affiliations
  • 1 School of Electrical Automation and Information Engineering, Tianjin University, Tianjin 300072, China
  • 2 School of Microelectronics, Tianjin University, Tianjin 300072, China
    • show less
    AbstractGet PDF(in Chinese)
    Figures&Tables (14)
    Equations (0)
    References (0)
    Tools
    Paper Information
    Topics
    Figures & Tables(14)
    Architecture of region proposal network

    Fig. 1. Architecture of region proposal network

    Download full size
    Multi-view samples. (a) Frontal view; (b) profile view; (c) back view

    Fig. 2. Multi-view samples. (a) Frontal view; (b) profile view; (c) back view

    Download full size
    Architecture of MVNN

    Fig. 3. Architecture of MVNN

    Download full size
    Principal detection neural network model of MVNN

    Fig. 4. Principal detection neural network model of MVNN

    Download full size
    Three channels of input layer. (a) Sample 1; (b) sample 2

    Fig. 5. Three channels of input layer. (a) Sample 1; (b) sample 2

    Download full size
    Calculation model of part score for deformation layer

    Fig. 6. Calculation model of part score for deformation layer

    Download full size
    Comparison results of region proposal for input data. (a) HOG+Adaboost algorithm; (b) Proposed region proposal algorithm

    Fig. 7. Comparison results of region proposal for input data. (a) HOG+Adaboost algorithm; (b) Proposed region proposal algorithm

    Download full size
    Testing result of multi-view model. (a) Testing result of multiple views; (b) Comparison results of single-view model and multi-view model

    Fig. 8. Testing result of multi-view model. (a) Testing result of multiple views; (b) Comparison results of single-view model and multi-view model

    Download full size
    Testing result of DPM

    Fig. 9. Testing result of DPM

    Download full size
    Testing result of proposed algorithm on IHDD. (a) RFPPI-RMR curve; (b) P-R curve

    Fig. 10. Testing result of proposed algorithm on IHDD. (a) RFPPI-RMR curve; (b) P-R curve

    Download full size

    • Table 1. Parameters of part filters in proposed algorithm

      View table

      Table 1. Parameters of part filters in proposed algorithm

      Parameter123456789101112131415
      Starting line114411141111-111
      Ending line339939993999999
      Starting column131311411114111
      Ending column353552555525555

    • Table 2. Indoor human detection dataset

      View table

      Table 2. Indoor human detection dataset

      Test environmentTotal framesAnnotated humansNo.
      Office day 139125163
      Office day 231574785
      Office day 3246394
      Duty room day 1108178
      Duty room day 2813893
      Duty room night 160636072
      Duty room night 2369369
      Training set879910479
      Validation set29333701
      Test set29333674
      Total number of samples1466517854

    • Table 3. Parameter setting of second-stage network model

      View table

      Table 3. Parameter setting of second-stage network model

      ParameterEpoch 1-150Epoch 151-250
      Learning rate0.0250.0125
      Momentum0.90.9
      Batch size8080

    • Table 4. Quantitative comparison of different algorithms on dataset

      View table

      Table 4. Quantitative comparison of different algorithms on dataset

      AlgorithmRMRRAP /%
      measurements(RFPPI-RMR) /%
      Ref. [3]37.3857.32
      Ref. [21]17.2984.82
      Ref. [18]28.8475.52
      Proposed14.6687.34
    Tools

    Get Citation

    Copy Citation Text

    Xia Wang, Wei Zhang. Multi-View Indoor Human Detection Neural Network Based on Joint Learning[J]. Acta Optica Sinica, 2019, 39(2): 0210002

    Download Citation

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

    Category: Image Processing

    Received: Aug. 6, 2018

    Accepted: Sep. 17, 2018

    Published Online: May. 10, 2019

    The Author Email:

    DOI:10.3788/AOS201939.0210002

    Topics

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