M3EL links multimodal mentions to knowledge base entities using intra-modal contrastive learning and bidirectional text-image matching, and reports state-of-the-art results on WikiMEL, RichpediaMEL and WikiDiverse.
Title resolution pending
1 Pith paper cite this work. Polarity classification is still indexing.
1
Pith paper citing it
fields
cs.CV 1years
2024 1verdicts
CONDITIONAL 1representative citing papers
citing papers explorer
-
Multi-level Matching Network for Multimodal Entity Linking
M3EL links multimodal mentions to knowledge base entities using intra-modal contrastive learning and bidirectional text-image matching, and reports state-of-the-art results on WikiMEL, RichpediaMEL and WikiDiverse.