PIGEONS: Incremental Long-Video Captioning with Persistent Identity Gallery and Evolving Overarching Narrative Summary
cs.CV
Submitted: 2026-10-12
Updated: 2026-10-12
Project page: https://pigeons-captioner.github.io/ABSTRACT
License: http://arxiv.org/licenses/nonexclusive-distrib/1.0/
Sources
- DiaDem: Advancing Dialogue Descriptions in Audiovisual Video Captioning for Multimodal Large Language Models
- CapRiCorn-1K: A Comprehensive Benchmark for Video Captioning and Subject Referential Consistency Across Temporal Scales
- Fine-Grained Captioning of Long Videos through Scene Graph Consolidation
- Gemini 2.5: Pushing the Frontier with Advanced Reasoning, Multimodality, Long Context, and Next Generation Agentic Capabilities
- ARC-Hunyuan-Video-7B: Structured Video Comprehension of Real-World Shorts
- StoryTeller: Improving Long Video Description through Global Audio-Visual Character Identification
- VideoChat-R1: Enhancing Spatio-Temporal Perception via Reinforcement Fine-Tuning
- Towards Universal Video MLLMs with Attribute-Structured and Quality-Verified Instructions
- ReBind: Multi-Reference Video Editing via Structured Instructions with Explicit Reference Relationships
- VideoCap-R1: Enhancing MLLMs for Video Captioning via Structured Thinking
- OmniScript: Towards Audio-Visual Script Generation for Long-Form Cinematic Video
- CinePile: A Long Video Question Answering Dataset and Benchmark
- Proximal Policy Optimization Algorithms
- DeepSeekMath: Pushing the Limits of Mathematical Reasoning in Open Language Models
- DINOv3
- video-SALMONN 2: Caption-Enhanced Audio-Visual Large Language Models
- LongCat-Flash-Omni Technical Report
- Script-a-Video: Deep Structured Audio-visual Captions via Factorized Streams and Relational Grounding
- Tarsier: Recipes for Training and Evaluating Large Video Description Models
- LongCaptioning: Unlocking the Power of Long Video Caption Generation in Large Multimodal Models
Related papers
- Loss Knows Best: Detecting Annotation Errors in Videos via Loss Trajectories
- AnchorWeave: World-Consistent Video Generation with Retrieved Local Spatial Memories
- Benchmarking the Robustness of Foundation Models for Mammography under Domain Shift
- MambaX-Net: Dual-Input Mamba-Enhanced Cross-Attention Network for Longitudinal MRI Segmentation
- TeleOCR: Navigating Document Parsing Across Digital and Camera-Captured Documents
- A Survey on Efficient Vision-Language-Action Models