• Laser & Optoelectronics Progress
  • Vol. 60, Issue 10, 1010019 (2023)
Cheng Liu1, Liangcai Cao2, Ye Jin3, Haowei Wang3, and Songfeng Yin1、*
Author Affiliations
  • 1Hefei Institute for Public Safety Research, Tsinghua University, Hefei 230601, Anhui , China
  • 2State Key Laboratory of Precision Measurement Technology and Instruments, Tsinghua University, Beijing 100084, China
  • 3Criminal Police Detachment of Hefei Public Security Bureau, Hefei 230601, Anhui , China
  • show less
    DOI: 10.3788/LOP220785 Cite this Article Set citation alerts
    Cheng Liu, Liangcai Cao, Ye Jin, Haowei Wang, Songfeng Yin. Transformer for Age-Invariant Face Recognition[J]. Laser & Optoelectronics Progress, 2023, 60(10): 1010019 Copy Citation Text show less
    Architecture of the T2T-DAL
    Fig. 1. Architecture of the T2T-DAL
    Network structure of face feature extraction model
    Fig. 2. Network structure of face feature extraction model
    RFM architecture
    Fig. 3. RFM architecture
    DatasetNumber of identitiesNumber of images
    faces emore11857425774205
    AgeDB-30136000 pairs12000
    CACD_VS144000 pairs8000
    CALFW156000 pairs12000
    LFW166000 pairs12000
    Table 1. Information about the used dataset
    ParameterProposed modelDAL(OE-CNNs)4MTLFace7
    MACs /GB1.28.156.32
    Params /10640.7157.0847.25
    Inference speed /msBatch size is 113.7525.0121.99
    Batch size is 161.192.692.54
    Table 2. Comparison of parameters of different models
    MethodAgeDB-3013CACD_VS14CALFW15LFW16
    DAL(OE-CNNs)499.499.47
    DAL(OE-CNNs)-Re95.9899.3895.4899.56
    MTLFace796.2399.5595.6299.52
    Proposed method96.2599.3595.4899.63
    Table 3. Accuracy of different methods on four datasets
    Cheng Liu, Liangcai Cao, Ye Jin, Haowei Wang, Songfeng Yin. Transformer for Age-Invariant Face Recognition[J]. Laser & Optoelectronics Progress, 2023, 60(10): 1010019
    Download Citation