Computer VisionFeatured briefing
First-hand information for everyone
A new preprint (arXiv:2610.21902v1) paints a careful picture of a surprising weakness in current multimodal large language models (MLLMs): l
A team from the University of Oxford has introduced an automated pipeline that can map the branching network of blood vessels in retinal pho
A team of researchers proposes a new way to teach game-playing agents that leans on synthetic practice rather than only on human demonstrati
A new preprint on arXiv raises fresh privacy concerns about a growing practice in medical AI: sharing learned representations of medical ima
Roboticists have long hoped for generalist policies—single controllers that can perform many different tasks—because they promise to shortcu
Researchers have introduced OccludeBench, a new benchmark that probes whether vision-language models (VLMs) can reason about objects that ar
A team of researchers has won first place in the ECCV 2026 LifeGenIP Challenge by developing a practical method to make short facial videos
A new method called READ offers a faster, simpler way to attribute predictions made by a growing class of video models, according to a paper
Computer-aided detection and diagnosis tools for breast cancer screening can help radiologists, but one stubborn problem is false negatives
A new preprint on arXiv examines whether training multimodal large language models (MLLMs) on in‑domain industrial inspection data is enough