TY - GEN
T1 - Vision-based gender recognition using hybrid background subtraction technique
AU - Takhar, Gourav
AU - Prakash, Chandra
AU - Mittal, Namita
AU - Kumar, Rajesh
N1 - Publisher Copyright:
© Springer Nature Singapore Pte Ltd. 2018.
PY - 2018
Y1 - 2018
N2 - Gait-Based Gender Classification (GBGC) is a relatively new field in gender classification based applications. Lots of work has been done on gender classification using voice and face. However, little, on the effect of background subtraction on gender classification. This paper focuses on analyzing the effects of background subtraction techniques used to obtained gait energy for GBGC and consider cases where the subject is injured or changes walking behavior intentionally. No prior research has been done on datasets containing walking behavior and effect of background subtraction on GBGC. ViMO and MOG2 are used as background subtraction techniques and applied on the dataset collected at MNIT- Jaipur containing 50 subjects (17 Female and 33 Male) with a total of 590 video sequences. The selected video sequences contained normal walk and unhealthy walk (both left and right) pattern. This paper shows ViMO technique performs better than state of the art MOG2 technique and effect of changing walking behavior is negligible.
AB - Gait-Based Gender Classification (GBGC) is a relatively new field in gender classification based applications. Lots of work has been done on gender classification using voice and face. However, little, on the effect of background subtraction on gender classification. This paper focuses on analyzing the effects of background subtraction techniques used to obtained gait energy for GBGC and consider cases where the subject is injured or changes walking behavior intentionally. No prior research has been done on datasets containing walking behavior and effect of background subtraction on GBGC. ViMO and MOG2 are used as background subtraction techniques and applied on the dataset collected at MNIT- Jaipur containing 50 subjects (17 Female and 33 Male) with a total of 590 video sequences. The selected video sequences contained normal walk and unhealthy walk (both left and right) pattern. This paper shows ViMO technique performs better than state of the art MOG2 technique and effect of changing walking behavior is negligible.
KW - Background subtraction
KW - Gait based gender classification (GBGC)
KW - MOG2
KW - ViMO
UR - http://www.scopus.com/inward/record.url?scp=85049011770&partnerID=8YFLogxK
U2 - 10.1007/978-981-10-8660-1_49
DO - 10.1007/978-981-10-8660-1_49
M3 - Conference contribution
AN - SCOPUS:85049011770
SN - 9789811086595
T3 - Communications in Computer and Information Science
SP - 651
EP - 662
BT - Smart and Innovative Trends in Next Generation Computing Technologies - 3rd International Conference, NGCT 2017, Revised Selected Papers
A2 - Bhattacharyya, Pushpak
A2 - Sastry, Hanumat G.
A2 - Marriboyina, Venkatadri
A2 - Sharma, Rashmi
PB - Springer Verlag
T2 - 3rd International Conference on Next Generation Computing Technologies, NGCT 2017
Y2 - 30 October 2017 through 31 October 2017
ER -