Pith. sign in

REVIEW 8 cited by

From Seconds to Hours: Reviewing MultiModal Large Language Models on Comprehensive Long Video Understanding

Not yet reviewed by Pith; the record is open.

This paper has not been read by Pith yet. Machine review is queued; the pith claim, tier, and objections will appear here once it completes.

SPECIMEN: schema-true, not a live event

T0 review · schema-true

One-sentence machine reading of the paper's core claim.

pith:XXXXXXXX · record.json · timestamp

arxiv 2409.18938 v2 pith:TOOJDVFI submitted 2024-09-27 cs.CV cs.AI

From Seconds to Hours: Reviewing MultiModal Large Language Models on Comprehensive Long Video Understanding

classification cs.CV cs.AI
keywords understandinglongvideomm-llmsvideosvisuallanguagelarge
verification ladder T0 review T1 audit T2 compute T3 formal T4 reserved
0 comments
Share X Bluesky LinkedIn Reddit HN
read the original abstract

The integration of Large Language Models (LLMs) with visual encoders has recently shown promising performance in visual understanding tasks, leveraging their inherent capability to comprehend and generate human-like text for visual reasoning. Given the diverse nature of visual data, MultiModal Large Language Models (MM-LLMs) exhibit variations in model designing and training for understanding images, short videos, and long videos. Our paper focuses on the substantial differences and unique challenges posed by long video understanding compared to static image and short video understanding. Unlike static images, short videos encompass sequential frames with both spatial and within-event temporal information, while long videos consist of multiple events with between-event and long-term temporal information. In this survey, we aim to trace and summarize the advancements of MM-LLMs from image understanding to long video understanding. We review the differences among various visual understanding tasks and highlight the challenges in long video understanding, including more fine-grained spatiotemporal details, dynamic events, and long-term dependencies. We then provide a detailed summary of the advancements in MM-LLMs in terms of model design and training methodologies for understanding long videos. Finally, we compare the performance of existing MM-LLMs on video understanding benchmarks of various lengths and discuss potential future directions for MM-LLMs in long video understanding.

discussion (0)

Sign in with ORCID, Apple, or X to comment. Anyone can read and Pith papers without signing in.

Forward citations

Cited by 8 Pith papers

Reviewed papers in the Pith corpus that reference this work. Sorted by Pith novelty score.

  1. TennisTV: Do Multimodal Large Language Models Understand Tennis Rallies?

    cs.CV 2025-09 unverdicted novelty 7.0

    Introduces TennisTV benchmark for evaluating 17 MLLMs on tennis video understanding from stroke-level to rally-level tasks with automated pipelines and human verification.

  2. Homer: Understanding Long-form Videos with Hierarchical Memory and Agentic Reasoning

    cs.CV 2026-07 accept novelty 6.5

    Hierarchical online memory with explicit temporal-causal event edges plus a verify-and-correct agentic reasoner yields large gains on long-form video QA in the streaming setting.

  3. Empowering Long-form Omni-modal Understanding with Robust Audio Perception

    cs.LG 2026-07 conditional novelty 6.0

    Decoupled audio-visual caption and CoT-QA datasets plus two-stage fine-tuning measurably strengthen auditory perception and cross-modal reasoning in a 7B omni-modal LLM.

  4. Event-Causal RAG: A Retrieval-Augmented Generation Framework for Long Video Reasoning in Complex Scenarios

    cs.AI 2026-05 unverdicted novelty 6.0

    Event-Causal RAG segments videos into events represented as SES graphs, merges them into a causal knowledge graph, and uses bidirectional retrieval to supply relevant event chains to a video foundation model for impro...

  5. Reinforce to Learn, Elect to Reason: A Dual Paradigm for Video Reasoning

    cs.CV 2026-04 unverdicted novelty 6.0

    RLER trains video-reasoning models with three task-driven RL rewards for evidence production and elects the best answer from a few candidates via evidence consistency scoring, yielding 6.3% average gains on eight benchmarks.

  6. OOWM: Structuring Embodied Reasoning and Planning via Object-Oriented Programmatic World Modeling

    cs.AI 2026-02 unverdicted novelty 6.0

    OOWM models the world as an explicit symbolic tuple with UML diagrams and trains via SFT plus GRPO to outperform text-based CoT on embodied planning benchmarks.

  7. Training-free Uncertainty Guidance for Complex Visual Tasks with MLLMs

    cs.CV 2025-10 conditional novelty 6.0

    Selecting the visual input that minimizes an MLLM's output entropy (or maximizes its yes/no confidence) improves fine-grained visual search, long-video QA, and temporal grounding without any training.

  8. NeMo: Needle in a Montage for Video-Language Understanding

    cs.CV 2025-09 conditional novelty 5.0

    NeMoBench, an automatically generated benchmark with 31,378 QA pairs, shows that video LLMs struggle with temporal grounding of relevant clips hidden in long montages.