Hand-Based Gender Classification

Background

In computer vision, the majority of studies on gender classification are based on face because visual information from human faces provides important cues for gender classification. Less popular for gender classification is the use of modalities other than the face such as hands, irises and fingerprints, but using different parts of the body in gender classification can provide many advantages compared to the face.

Description

Our researchers at the University of Nevada, Reno have created a new method and system for performing hand-based gender classification by computing a set of feature parameters representing a geometry of the digitally-imaged hand part. Hand-based gender classification can be used solely for determining the gender of one or more subjects, or as a tool to improve other methods and systems such as bioinformatics systems and the computer’s perceptional and interactional capabilities.

Advantages

• Images of hands can be captured more robustly than images of face.
• Our classification system can improve intelligence of a surveillance system, collect valuable consumer statistics in a shopping center, and analyze customer demands for store management.
• Hand appearance shows less variability compared to face appearance (e.g., face appearance is affected by factors such as facial expression change, rest, hair growth and makeup).
• Hand-based gender classification could be used to improve the performance of other biometric systems, such as face-based authentication and identification/recognition systems.

Related Documents

• Patent 8,655,084: Hand-Based Gender Classification
• Similar technology we have: Hand-Based Biometric Analysis
• Patent 9,042,606: Hand-Based Biometric Analysis

UNR09-019