On covariate factor detection and removal for robust gait recognition

Abstract

We propose a novel bolt-on module capable of boosting the robustness of various single compact 2D gait representations. Gait recognition is negatively influenced by covariate factors including clothing and time which alter the natural gait appearance and motion. Contrary to traditional gait recognition, our bolt-on module remedies this by a dedicated covariate factor detection and removal procedure which we quantitatively and qualitatively evaluate. The fundamental concept of the bolt-on module is founded on exploiting the pixel-wise composition of covariate factors. Results demonstrate how our bolt-on module is a powerful component leading to significant improvements across gait representations and datasets yielding state-of-the-art results.

Robust gait recognition: a comprehensive survey

“…Another trend is to estimate the position of covariates in order to remove them. Whytock et al [103] proposed a novel bolt-on module enabling to improve the robustness using various single compact 2D gait representations including GEI. Recently, Ghebleh and Ebrahimi [110] introduced an adaptive outlier detection method to address the effects of clothing issue.…”

“…Zhang et al [123] [135] introduced the concept of accumulated flow image and edge-masked active energy image able to produce distinctive features for classification. Lee et al [136] proposed a combination of spatio-temporal approach and texture descriptors to extract discriminative gait features named [88] 2010 anatomical properties • Choudhury and Tjahjadi, [89] 2015 anatomical properties • Verlekar et al [90] 2017 anatomical properties • Aggarwal and Vishwakarma [91] 2017 anatomical properties • Li and Chen [92] 2013 self-defined • Iwashita et al [93] 2013 self-defined • Gabriel et al [94] 2013 self-defined • Islam et al [95] 2013 self-defined • Nandy et al [96] 2016 self-defined • Lishani et al [97] 2017 self-defined • Bashir et al [98] 2008 wrapper • Dupuis et al [99] 2013 random forest • Rida et al [100] 2014 wrapper • Rida et al [101] 2015 wrapper • Rokanujjaman et al [102] 2015 wrapper • Whytock et al [103] 2015 bolt-on module • Rida et al [104] 2016 wrapper • Rida et al [105,106] 2016 group fused Lasso • Rida et al [78] 2016 SD • Alotaibi and Mahmood [107,108] 2016 Gini impurity • Issac et al [109] 2017 genetic algorithm • Ghebleh and Ebrahimi [110] 2017 adaptive outlier detection • Liang et al [111] 2016 cloth proportion transient binary patterns. Lee et al [137] applied HOG to timesliced averaged motion history image in order to extract discriminative features.…”

Gait recognition from corrupted silhouettes: a robust statistical approach

“…Finally, each original silhouette is substituted by a new image (FDEI) that arises from the summation of its cluster's DEI and the positive portion of its difference with the preceding silhouette. A different strategy to deal with problems in silhouettes is to exclude affected regions as in [30], where covariate factors are removed from gait representations.…”

Análisis automático del patrón de la marcha mediante visión por computador: problemas, métodos y aplicaciones

“…Finally, each original silhouette is substituted by a new image (FDEI) that arises from the summation of its cluster's DEI and the positive portion of its difference with the preceding silhouette. A different strategy to deal with problems in silhouettes is to exclude affected regions as in [99], where covariate factors are removed from gait representations.…”