Video Captioning
Overview
Video captioning studies how to describe what happens in a clip, from a single concise sentence to dense, timestamped event descriptions and long-form narratives. Systems combine video-language models with temporal segmentation and grounding, frame selection, tracking or motion cues, and—in audiovisual settings—alignment between sounds, speakers, and visible events. Common test settings include short clips and instructional videos, using benchmarks such as MSR-VTT, VATEX, ActivityNet Captions, and YouCook2; newer evaluations also test factuality, temporal accuracy, entity consistency, and whether captions support downstream question answering. A central challenge is balancing completeness and fluent detail against hallucination, especially as videos grow longer, while standard text-overlap metrics often miss these qualities.
Latest papers
71 papers7 papers in the last four weeks, level with the four weeks before. 0.1% of all new papers.
PhysCapBench: A Cross-Domain Benchmark for Physically Grounded Video Captioning
Reviso: Not All Visual Evidence Deserves Equal Attention in Video Captioning
CodeCap-Bench: Can Code Serve as a Good Video Caption?
MVRCap: Towards Comprehensive Video Captioning via Hierarchical Multi-Dimensional Verifiable Rewards
MarineLife-16K: An Attribute-Rich Video-Text Dataset for Motion-Aware Marine Video Understanding
StateBridge: Phase- and Track-Grounded Reasoning for Synthetic-to-Real Traffic Video Captioning and VQA
OACap: Caption-Guided Active Observation for Detailed Video Captioning
EVA-Cap: Optimizing Audiovisual Video Captioning via Event-Centric Alignment
PIGEONS: Incremental Long-Video Captioning with Persistent Identity Gallery and Evolving Overarching Narrative Summary
Beyond Anonymous Captions: Grounding Character Identity in Video Captioning and Question Answering
Physis-Lang: Self-Evolving Language as a Physical Representation for Video World Model
Comparative study of adapting pre-trained models for driving behavior video captioning
Causal-EVC: Breaking Emotional Spurious Causality via Spatiotemporal Grounding and Counterfactual Intervention
sadness'' as joy'' on a sunny beach), leading to severe shortcut learning from confusing backgrounds. Furthermore, existing evaluations fail to verify whether models have genuinely mastered causal reasoning or merely exploited background confounders. To address these limitations, we first construct {EVC-CauseGround}, a comprehensive benchmark with dense spatio-temporal causal annotations. Crucially, it introduces a carefully selected {Causal-Faithfulness Subset} to explicitly quantify genuine emotion-cause attribution. Second, we propose {Causal-EVC}, an emotion-grounding captioning framework, which introduces a Motion-guided Causal Spatiotemporal Localization module to precisely decouple causal triggers from background confounders. Besides, we introduce an Interpretable Sparse Emotion Routing module. By synthesizing counterfactual representations and formulating a novel counterfactual contrastive objective, we enforce the model to anchor its emotion predictions strictly on authentic causal triggers instead of confusing background. Extensive experiments show that Causal-EVC not only achieves the best performance on semantic metrics but also exhibits significant advantages in the causal-faithfulness subset, which demonstrates that our model could mine emotional cues from genuine visual causes and mitigate co-occurrence bias for interpretable multimodal emotion understanding.