Research Article

Gender Recognition from Unconstrained and Articulated Human Body

Table 5

Gender recognition accuracies (%) using the PLS method for dimensionality reduction (denoted by PLS_1) or both dimensionality reduction and classification (denoted by PLS_2) on PASCAL data with annotated body parts.

ā€‰ RLA LLA RUA LUA T H H + T H + T + LUA H + T + RUA H + T + RUA + LUA

HOG_PLS_1 58.90 58.90 62.38 60.28 58.90 65.23 65.21 65.97 70.11 69.56
LBP_PLS_1 58.90 58.90 62.30 60.06 62.59 65.25 67.28 67.36 69.93 69.01
RGB_PLS_1 57.37 58.91 58.70 58.12 57.73 58.52 58.90 58.90 58.20 58.90

HOG_PLS_2 52.82 48.89 56.85 56.53 57.02 62.04 59.64 58.05 58.59 58.73
LBP_PLS_2 50.98 50.69 54.95 53.00 57.84 63.09 65.40 66.69 69.41 68.72
RGB_PLS_2 53.27 51.64 58.58 50.66 58.27 56.04 52.66 53.97 54.10 53.84

HOG + LBP_1 58.90 58.90 63.10 61.59 63.19 67.09 66.40 68.14 70.57 69.08
HOG + LBP_2 48.07 50.50 55.70 54.08 60.25 66.11 67.67 67.59 69.80 69.57