Lighterr.com
AI-Assisted Product Design Data Labeling
Client:
lighterr.com
Scope:
Industrial Design
Year:
2024
Lighterr.com needed a structured product image dataset to improve its AI-driven product recommendations and visual search. The existing catalog contained inconsistently labeled images, duplicate entries, and unclear categories — limiting the accuracy of their recommendation engine.
Our Role
xLabeling developed a human-in-the-loop image labeling process to standardize and classify Lighterr’s product images across multiple categories, models, and color variations.
Process
Reviewed over 5,000 product images from Lighterr’s online catalog.
Defined a clear labeling taxonomy (brand, model, finish, use case).
Annotated each image with standardized tags and verified visual consistency.
Conducted dual-review QA to ensure accuracy and eliminate duplicates.
Result
The final labeled dataset enabled Lighterr.com to deploy a visual search feature and improve product recommendations, increasing catalog accuracy by 98% and reducing manual data cleanup time by 60%.