A new arXiv paper introduces a dataset for evaluating how well large language and multimodal models can analyze multi-panel scientific charts through visual question answering across four levels of…

#AIresearch #MultimodalModels #VQA #DigitalLibraries
https://arxiv.org/abs/2610.14173

ILCD model advances crop disease diagnosis via visual question answering, integrating coattention, MUTAN, and bias-balancing to identify disease stages and attributes accurately. #CropDisease #VQA #AgTech
Details: doi.org/10.34133/pla...

Small vision-language models often read external evidence well but struggle to retrieve it. Harness Compilation adapts the division of work between a frozen small VLM and its harness using teacher traces,…

#HarnessCompilation #VisionLanguageModels #VQA #SmallVLMs
https://arxiv.org/abs/2610.11231

Is the wine scorecard dead?

The old signals of wine quality might not be as important to a younger generation of wine customers. How to market your winery to the next generation of consumers.

www.mikebelobradic.com/blog/is-the-...

#winebusiness #winerymarketing #winemarketing #winecountry #VQA

Organisers: Viorica Patraucean, Joao Carreira, Fedor Kitashov, Shiry Ginosar, Dima Damen, Andrew Zisserman

#ECCV #PerceptionTest #SpatialAI #ComputerVision #VQA #MachineLearning