TVQA

TVQA is a dataset that contains video clips paired with questions and answers, used to evaluate the performance of models in video question answering tasks.

Papers using TVQA (10)

Multi-speaker Attention Alignment for Multimodal Social Interaction2025POVQA: Preference-optimized Video Question Answering With Rationales For Data Efficiency2025Multimodal Conversation Structure Understanding2025Zero-Shot Video Question Answering via Frozen Bidirectional Language Models2022 · 64 citesVideo Question Generation via Cross-Modal Self-Attention Networks Learning2019 · 2 citesModality Shifting Attention Network for Multi-modal Video Question Answering2020 · 1 citesFrame-Subtitle Self-Supervision for Multi-Modal Video Question Answering2022 · 1 citesMERLOT Reserve: Neural Script Knowledge through Vision and Language and Sound2022TV-TREES: Multimodal Entailment Trees for Neuro-Symbolic Video Reasoning2024On Modality Bias in the TVQA Dataset2020