AI Packaging Artwork Parser for Multimodal Document Intelligence
Innovation prototype for extracting searchable structure from complex packaging artwork.
Context
Innovation Days proof-of-concept for packaging artwork understanding.
Problem
Packaging artwork files contain complex combinations of text, visuals, layout, and image-based information that are difficult to search and interpret manually.
My Role
Contributed to AI solution design and implementation for extracting, structuring, searching, and interacting with parsed artwork data.
Impact / Output
Recognized as a final-five Innovation Days project and third crowd favourite among 59 teams.
Technical Approach
Document intelligence, vision-transformer concepts, image/PDF parsing, Cosmos DB, AI search/chat layer, and multimodal information extraction.