A data-driven classification of 3D foot types by archetypal shapes based on landmarks

Area of Science:

• Anthropometry
• Biomechanical Engineering
• Data Science

Background:

• Establishing accurate taxonomies of body shapes, particularly foot morphology, is crucial for effective design applications like footwear.
• Existing methods for creating typologies often have limitations, hindering objective and data-driven classification.

Purpose of the Study:

• To introduce and validate a novel methodology, Archetypoid Analysis (ADA), for objective and data-driven taxonomy creation.
• To overcome the limitations of traditional clustering techniques in identifying extreme patterns within datasets.
• To apply ADA to a 3D foot scan database for practical applications in footwear design.

Main Methods:

• Archetypoid Analysis (ADA), an objective, data-driven approach focusing on identifying extreme patterns (archetypal profiles) within data.
• Application of ADA to a database of 775 3D right foot scans from the Spanish adult population.
• Analysis of foot shapes described by high-dimensional landmark data (5626x3 configuration matrix) without using pre-defined multivariate features.

Main Results:

• Successful identification of 3 distinct archetypal foot shapes for both adult males and females.
• Demonstration that ADA provides an understandable explanation of data as percentages of archetypal patterns.
• Evidence that ADA can uncover extreme patterns not discernible through conventional multivariate or clustering techniques.

Conclusions:

• Archetypoid Analysis (ADA) offers a robust and accessible method for establishing objective taxonomies from complex shape data.
• The identified archetypal feet provide valuable insights for targeted footwear design and customization.
• ADA's ability to identify extreme patterns offers advantages over traditional methods for shape analysis and classification.