How Hugo's Human Expertise Enhanced 3D Foot Scanning Technology

Overview

The Client helps millions of shoppers find the perfect shoe fit by combining 3D foot scans, retail purchase data, and AI-driven insights. To enhance their mobile foot scanning app—an essential part of their technology suite, the Client partnered with Hugo for high-precision data labeling.

This collaboration enabled retailers and brands to deliver more accurate, frictionless, and personalized shopping experiences both in-store and online.

Challenges

Even with state-of-the-art sensors, achieving accurate foot measurements is complex. Existing AR frameworks lacked the precision needed for the Client’s mobile foot scanner, leading them to develop proprietary models. To achieve the highest accuracy, they needed:

• Pixel-perfect labeling: Every pixel impacts measurement accuracy. The Client required precise raster masks to ensure their computer vision (CV) models captured even the smallest details.

• Handling diverse edge cases: Shadows, low-contrast lighting, and occlusions posed challenges for AI interpretation. Data labelers needed to adapt to these complexities to improve model performance.

The Client sought a partner capable of refining models through tight feedback loops while also scaling quickly to meet increasing labeling demands.

Solution

Hugo addressed these challenges by combining expert human annotation with rigorous quality control:

Ensuring Pixel-Perfect Precision:

• Hugo assembled a team with an anatomical background to deliver precise labeling, even in ambiguous cases.

• Hugo implemented a multi-stage human review process. An initial annotator created the mask, a senior annotator reviewed and refined it, and a quality lead conducted final verification. This approach caught errors automated systems might miss.

• To ensure consistency, Hugo developed reference materials outlining protocols for boundary scenarios with visual examples. Annotators participated in weekly calibration sessions to review difficult cases and align on best practices, maintaining accuracy in foot scanning.

Precision Annotation for Complex Edge Cases

• Hugo tackled challenging scenarios by analyzing historical scan data, identifying 30+ problematic cases, and creating tailored guidelines.

• For shadow complications, trained annotators used texture and gradient cues to distinguish foot boundaries, especially for lower-contrast skin tones. For occlusions, anatomically trained specialists predicted covered sections using visible landmarks. In low-contrast cases, experienced analysts relied on reference libraries to identify subtle boundaries

• Difficult cases were reviewed through a consensus approach, where multiple annotators labeled the same image before a senior specialist resolved discrepancies. To ensure complex cases were handled effectively, Hugo matched scans to annotators with relevant expertise.

Scalable Workforce Without Compromising Quality

• Hugo introduced a tiered certification program, enabling new annotators to manage simpler tasks while experienced team members handled complex cases.

• To uphold quality, project managers conducted regular audits, tracking accuracy and productivity. 

• Hugo adapted to evolving requirements through direct communication with the Client’s engineering team in twice-weekly meetings. Training specialists updated guidelines and conducted focused training as needed.

• For testing new measurement approaches, Hugo assigned a specialized team to implement experimental protocols without disrupting ongoing operations.

The Results

Hugo's partnership delivered measurable improvements:

• Achieved an average annotation Quality Score of 97.84%

• Maintained continuous feedback loops to refine model performance and address edge cases proactively

• Ensured accurate dataset diversity, resulting in improved and hyper-personalized shoe recommendations for customers worldwide.

"Having worked with different Vendors where the staff doing the actual work was always very hidden from us, we appreciated the transparency and social sustainability of Hugo."
— James A. Sr. Head of Engineering