Gait, the walking pattern of individuals, is one of the most important biometrics modalities. Most of the existing gait recognition methods take silhouettes or articulated body models as the gait features. These methods suffer from degraded recognition performance when handling confounding variables, such as clothing, carrying and view angle. To remedy this issue, we propose a novel AutoEncoder framework to explicitly disentangle pose and appearance features from RGB imagery and the LSTM-based integration of pose features over time produces the gait feature. In addition, we collect a Frontal-View Gait (FVG) dataset to focus on gait recognition from frontal-view walking, which is a challenging problem since it contains minimal gait cues compared to other views. FVG also includes other important variations, e.g., walking speed, carrying, and clothing. With extensive experiments on CASIA-B, USF and FVG datasets, our method demonstrates superior performance to the state of the arts quantitatively, the ability of feature disentanglement qualitatively, and promising computational efficiency.

Overview Gait Net

Figure 2: Overall architecture of our proposed approach, with three novel loss functions.

Overview FVG Dataset

Figure 3: Examples of FVG Dataset. (a) Samples of the near frontal middle, left and right walking view angles in session 1 (SE1) of the first subject (S1). SE3-S1 is the same subject in session 3. (b) Samples of slow and fast walking speed for another subject in session 1. Frames in top red boxes are slow and in the bottom red box are fast walking. Carrying bag sample is shown below. (c) samples of changing clothes and with cluttered background from one subject in session 2.

FVG Database

FVG database download page: http://cvlab.cse.msu.edu/frontal-view-gaitfvg-database.html

Publications

  • On Learning Disentangled Representations for Gait Recognition
    Ziyuan Zhang, Luan Tran, Feng Liu, Xiaoming Liu
    IEEE Transactions on Pattern Analysis and Machine Intelligence, , Sep. 2019 (Under Review)
    Bibtex | PDF | arXiv
  • @article{ on-learning-disentangled-representations-for-gait-recognition,
      author = { Ziyuan Zhang and Luan Tran and Feng Liu and Xiaoming Liu },
      title = { On Learning Disentangled Representations for Gait Recognition },
      booktitle = { IEEE Transactions on Pattern Analysis and Machine Intelligence },
      month = { September },
      year = { 2019 },
    }
  • Gait Recognition via Disentangled Representation Learning
    Ziyuan Zhang, Luan Tran, Xi Yin, Yousef Atoum, Jian Wan, Nanxin Wang, Xiaoming Liu
    In Proceeding of IEEE Computer Vision and Pattern Recognition (CVPR 2019), Long Beach, CA, Jun. 2019 (Oral presentation)
    Bibtex | PDF | arXiv
  • @inproceedings{ gait-recognition-via-disentangled-representation-learning,
      author = { Ziyuan Zhang and Luan Tran and Xi Yin and Yousef Atoum and Jian Wan and Nanxin Wang and Xiaoming Liu },
      title = { Gait Recognition via Disentangled Representation Learning },
      booktitle = { In Proceeding of IEEE Computer Vision and Pattern Recognition },
      address = { Long Beach, CA },
      month = { June },
      year = { 2019 },
    }