wovepaper — the arXiv observatory

The paper presents VETO, a subtle anti-edit cloak that disrupts how modern diffusion-based image editors read source images, and introduces VetoBench, a benchmark for evaluating pr…

Publications (88)

When To Solve, When To Verify: Compute-Optimal Problem Solving and Generative Verification for LLM Reasoning

Multimodal Explanations: Justifying Decisions and Pointing to the Evidence

Visual Coreference Resolution in Visual Dialog using Neural Module Networks

Learning to Reason: End-to-End Module Networks for Visual Question Answering

KRISP: Integrating Implicit and Symbolic Knowledge for Open-Domain Knowledge-Based VQA

Recognizing Fine-Grained and Composite Activities using Hand-Centric Features and Script Data

Learning to Generate Grounded Visual Captions without Localization Supervision

Coherent Multi-Sentence Video Description with Variable Level of Detail

DEFAME: Dynamic Evidence-based FAct-checking with Multimodal Experts

Obliviate: Erasing Concepts from Autoregressive Image Generation Models

VETO: Towards Protecting Images From Frontier AI Editing

A Dataset for Movie Description

Learning To Recognize Procedural Activities with Distant Supervision

Attentive Explanations: Justifying Decisions and Pointing to the Evidence

Decoupling Representation and Classifier for Long-Tailed Recognition

Movie Description

Spatial Semantic Regularisation for Large Scale Object Detection

Chrono: A Simple Blueprint for Representing Time in MLLMs

Attentive Explanations: Justifying Decisions and Pointing to the Evidence (Extended Abstract)

Grounding of Textual Phrases in Images by Reconstruction

Calibrated Sampling-Free Uncertainty Estimation in Bayesian Deep Learning

Reliable Visual Question Answering: Abstain Rather Than Answer Incorrectly

Improving Selective Visual Question Answering by Learning from Your Peers

A Multi-scale Multiple Instance Video Description Network

Deep Compositional Captioning: Describing Novel Object Categories without Paired Training Data

Erased but Not Forgotten: How Backdoors Compromise Concept Erasure

Uncertainty-guided Continual Learning with Bayesian Neural Networks

HaloProbe: Bayesian Detection and Mitigation of Object Hallucinations in Vision-Language Models

Sequence to Sequence -- Video to Text

The Long-Short Story of Movie Description

Geometric Erasure by Contrastive Velocity Matching in Rectified Flows

CLEVR-Dialog: A Diagnostic Dataset for Multi-Round Reasoning in Visual Dialog

Adversarial Inference for Multi-Sentence Video Description

VeriTaS: The First Dynamic Benchmark for Multimodal Automated Fact-Checking

Evaluating the Impact of Post-Training Quantization on Reliable VQA with Multimodal LLMs

12-in-1: Multi-Task Vision and Language Representation Learning

Token by Token, Compromised: Backdoor Vulnerabilities in Unified Autoregressive Models

Ask Your Neurons: A Deep Learning Approach to Visual Question Answering

Learning to Compose Neural Networks for Question Answering

Efficient Lifelong Learning with A-GEM

Iterative Answer Prediction with Pointer-Augmented Multimodal Transformers for TextVQA

Natural Language Object Retrieval

Simple Token-Level Confidence Improves Caption Correctness

On Tiny Episodic Memories in Continual Learning

Utilizing Large Scale Vision and Text Datasets for Image Segmentation from Referring Expressions

A New Split for Evaluating True Zero-Shot Action Recognition

Selfless Sequential Learning

Cycle-Consistency for Robust Visual Question Answering

Multimodal Compact Bilinear Pooling for Visual Question Answering and Visual Grounding

Generating Descriptions with Grounded and Co-Referenced People

Pythia v0.1: the Winning Entry to the VQA Challenge 2018

Drop an Octave: Reducing Spatial Redundancy in Convolutional Neural Networks with Octave Convolution

Think Twice, Act Once: Verifier-Guided Action Selection For Embodied Agents

In Defense of Grid Features for Visual Question Answering

TextCaps: a Dataset for Image Captioning with Reading Comprehension

CLASTER: Clustering with Reinforcement Learning for Zero-Shot Action Recognition

Captioning Images with Diverse Objects

Generating Visual Explanations

Memory Aware Synapses: Learning what (not) to forget

Large-Scale Visual Relationship Understanding

Translating Videos to Natural Language Using Deep Recurrent Neural Networks

ReCap: Lightweight Referential Grounding for Coherent Story Visualization

Speaking the Same Language: Matching Machine to Human Captions by Adversarial Training

Grounded Video Description

DMC-Net: Generating Discriminative Motion Cues for Fast Compressed Video Action Recognition

Ask Your Neurons: A Neural-based Approach to Answering Questions about Images

Neural Module Networks

Efficient Pre-training for Localized Instruction Generation of Videos

Graph-Based Global Reasoning Networks

Towards VQA Models That Can Read

Predicting Implicit Arguments in Procedural Video Instructions

SIEVES: Selective Prediction Generalizes through Visual Evidence Scoring

Attributes as Semantic Units between Natural Language and Visual Recognition

V$^2$Dial: Unification of Video and Visual Dialog via Multimodal Experts

Spurious-Aware Prototype Refinement for Reliable Out-of-Distribution Detection

Segmentation from Natural Language Expressions

FLAVA: A Foundational Language And Vision Alignment Model

CoDraw: Collaborative Drawing as a Testbed for Grounded Goal-driven Communication

Long-term Recurrent Convolutional Networks for Visual Recognition and Description