Investigating the use of motion-based features from optical flow for gait recognition

Gait biometrics: investigating the use of the lower inner regions for people identification from landmark frames

“…The obtained CCRs using the proposed method as well as for the other recent approaches using the KNN and deep learning classifiers being applied to the CASIA‐B data set are shown in Table 1. The results from other studies are reported directly from the work of Marfouf et al [25] who run the experiments using different descriptors on the same CASIA‐B data set which is used in this research. The majority of existing studies presented in Table 1 depend on the use of one complete gait cycle containing at least 22 successive frames.…”

“…The proposed inner lower limb region descriptor is evaluated using deep learning with the three‐fold cross‐validation. We have opted for the same framework of the deep learning procedure employed by Mahfouf et al [25] in order to conduct comparative analysis. To classify the extracted features by the scale conjugate gradient training algorithm, three superimposed auto‐encoders are utilised in tandem with a softmax network layer on top of size 100 in our case which represents the number of classes as shown in Fig.…”

“…11. The obtained results show that using only 14 frames, a better CCR is attained via the proposed approach compared to utilising 24 frames using two approaches LFAI [25] and GEI [15] which are 94.3 and 86.2%, respectively. The obtained CCR for 12 frames or less is also higher compared to both approaches LFAI and GEI.…”

“…Space‐time interest points are used to derive gait features at the temporal and spatial domains by taking into account the local variations at both domains. Mahfouf et al [25] estimated the optical frames across consecutive frames to introduce various gait descriptors for people identification. A number of empirical evaluations were conducted on the CASIA‐B gait database to assess motion‐based features estimated via optical flow to recognise people subjected to various factors.…”

An Appearance Invariant Gait Recognition Technique Using Dynamic Gait Features

“…In [24], Yu et al adapted optical flow field and histograms for gait recognition robust to appearance variance. Mahfouf et al [25] computed optical flow gait features for neural network-based gait recognition. In [26], Wang et al utilized gait silhouette as a set of three images and adapted a multichannel neural network.…”