Researchers have developed a new multimodal framework called SCENT that uses language guidance to bridge the gap between visual and olfactory information. This framework leverages Vision-Language Models (VLMs) to generate scene descriptors that capture objects, environmental context, and plausible smell cues, which then guide the learning of olfactory representations. Experiments on the New York Smells dataset show that SCENT significantly improves crossmodal retrieval tasks, outperforming vision-only baselines and achieving state-of-the-art results in smell-to-image and smell-to-text retrieval. AI
IMPACT This research could lead to more sophisticated AI systems capable of understanding and interpreting sensory data beyond vision, potentially impacting fields like robotics and environmental monitoring.
RANK_REASON The cluster contains an academic paper describing a new multimodal learning framework.
- alphaXiv
- arXiv
- CatalyzeX
- DagsHub
- Eleftherios Tsonis
- Gotit.pub
- Hugging Face
- New York Smells dataset
- ScienceCast
- Vision-Language Models
AI-generated summary · Google Gemini · from 2 sources. How we write summaries →
