Building a Large-scale Multimodal Knowledge Base System for Answering Visual Queries
This addresses the problem of deeper reasoning in visual queries for AI and computer vision researchers, offering a flexible alternative to purpose-built models, though it is incremental in combining existing data types into a scalable framework.
The authors tackled the challenge of comprehensive visual understanding by proposing a large-scale multimodal knowledge base system that handles diverse visual queries without training new classifiers, achieving competitive results on standard tasks with a scalable construction system that builds a KB with half a billion variables in hours.
The complexity of the visual world creates significant challenges for comprehensive visual understanding. In spite of recent successes in visual recognition, today's vision systems would still struggle to deal with visual queries that require a deeper reasoning. We propose a knowledge base (KB) framework to handle an assortment of visual queries, without the need to train new classifiers for new tasks. Building such a large-scale multimodal KB presents a major challenge of scalability. We cast a large-scale MRF into a KB representation, incorporating visual, textual and structured data, as well as their diverse relations. We introduce a scalable knowledge base construction system that is capable of building a KB with half billion variables and millions of parameters in a few hours. Our system achieves competitive results compared to purpose-built models on standard recognition and retrieval tasks, while exhibiting greater flexibility in answering richer visual queries.