AI Paper Atlas

Research map / Vision-language and multimodal / Multimodal LLMs

ViLU: Learning Vision-Language Uncertainties for Failure Prediction

Marc Lafon, Yannis Karmim, Julio Silva-Rodríguez, Paul Couairon, Clément Rambour, Raphael Fournier-Sniehotta, Ismail Ben Ayed, Jose Dolz, Nicolas Thome

ICCV 2025 · Vision-language and multimodal › Multimodal LLMs · approach: Bias-aware multi-image robustness · 5 citations (OpenAlex)

Read the paperCheck your idea against 137,217 papers

AI-extracted. The measured effects below were read by AI from the paper's full text (no summary yet). Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.

Measured effects reported in the paper

What was changedWhereMeasuredEffect
add predicted caption embedding to visual embedding (ablation)ViLU · ImageNet-1kAUC: 78.7 -> 88.8Improved
add cross-attention on top of visual embedding and predicted caption (ablation)ViLU · CC12MAUC: 88.9 -> 95.2Improved
add cross-attention (ablation)ViLU · CIFAR-10FPR95: 10.8 -> 8.2Improved
add predicted caption to model with cross-attention (ablation)ViLU · ImageNet-1kAUC: 86.1 -> 89.5Improved
train as binary classification instead of MSE loss regression (ablation)ViLU · ImageNet-1kAUC: 84.8 -> 88.6Improved
add automatic class weighting to BCE (ablation)ViLU · ImageNet-1kAUC: 88.6 -> 89.5Improved
ViLU instead of MCM on image-label datasetsViLU · 13 image-label datasets averageFPR95: 70.6 -> 29.9Improved
ViLU instead of BayesVLM on CC12MViLU · CC12MFPR95: 53.3 -> 25.2Improved
LVU vision-only loss prediction applied to image-caption datasetsLVU · CC12MFPR95: 58.8 -> 76.5Hurt
ViLU on low zero-shot accuracy datasetViLU · EuroSATAUC: 64.1 -> 98.8Improved
increase inference batch size from 128 to 32768ViLU · CC12MAUC: 96.9 -> 91.1Hurt
5-shot adaptation of ViLU on ImageNet-SketchViLU · ImageNet-SketchFPR95: 73.1 -> 54.4Improved

Later papers that compared with this method

1 later accepted paper reports ViLU as a baseline in their results tables; 1 of them report a better number on most of the results they share with it.

Matched by method name in results tables read by AI. Authors' own tables tend to favour the newer method, and a comparison can be missed when a method is renamed.

Closest papers by meaning

Writing a paper in this area? scime draft is a free LaTeX editor with AI included, real-time co-authors and a check of your references and venue rules before you submit. It also imports OpenAI Prism projects.

Open scime draft

Get the next scime Dispatch and a monthly field update

Free. We send a confirmation link first; every email has a one-click unsubscribe. Privacy