New York Smells dataset aims to give machines a sense of smell

Researchers have introduced "New York Smells," a novel multimodal dataset designed to bridge the gap between machine perception and the sense of smell. This dataset comprises 7,000 image-olfactory signal pairs, featuring over 3,500 distinct objects encountered in natural indoor and outdoor settings, significantly expanding upon existing olfactory datasets. Experiments using "New York Smells" demonstrate its utility in cross-modal olfactory representation learning, with learned representations outperforming traditional hand-crafted features on tasks such as smell-to-image retrieval and scene recognition. AI

IMPACT Enables new avenues for multimodal AI research by providing a large-scale dataset for olfactory perception.

RANK_REASON The item describes a new dataset and benchmark for olfaction research published on arXiv. [lever_c_demoted from research: ic=1 ai=1.0]

Read on arXiv cs.LG →

• alphaXiv
• arXiv
• Carl Vondrick
• CatalyzeX
• DagsHub
• Gotit.pub
• Hugging Face
• New York Smells
• ScienceCast

• paper
• other

AI-generated summary · Google Gemini · from 1 sources. How we write summaries →