Track papers across labs, topics, and trends.
Search papers, labs, and topics across Lattice.
5
0
7
3
MLLMs can recover missed food items and correct inaccurate portion estimations in single images, boosting nutrition estimation accuracy without task-specific fine-tuning.
Target-aware data selection and VLM-based caption refinement can boost food image retrieval performance by over 19%, revealing the potential of fine-tuning multimodal models with curated data.
Open-KNEAD achieves up to 53% better nutrition estimates than leading closed models, all while keeping user data private and local.
Ditch the depth sensors: DietDelta uses before-and-after photos and natural language to pinpoint exactly what you ate, item by item.
A new benchmark reveals how existing audio-visual segmentation models crumble when faced with the dynamic, ever-changing audio and visual environments of the real world.