Pith. sign in

REVIEW 24 cited by

SlowFast-LLaVA: A Strong Training-Free Baseline for Video Large Language Models

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 2407.15841 v2 pith:4C3XX3TH submitted 2024-07-22 cs.CV

SlowFast-LLaVA: A Strong Training-Free Baseline for Video Large Language Models

classification cs.CV
keywords videospatialfeaturesllmstraining-freecapturedesigndetailed
verification ladder T0 review T1 audit T2 compute T3 formal T4 reserved
0 comments
read the original abstract

We propose SlowFast-LLaVA (or SF-LLaVA for short), a training-free video large language model (LLM) that can jointly capture detailed spatial semantics and long-range temporal context without exceeding the token budget of commonly used LLMs. This is realized by using a two-stream SlowFast design of inputs for Video LLMs to aggregate features from sampled frames in an effective way. Specifically, the Slow pathway extracts features at a low frame rate while keeping as much spatial detail as possible (e.g., with 12x24 tokens), and the Fast pathway operates on a high frame rate but uses a larger spatial pooling stride (e.g., downsampling 6x) to focus on the motion cues. As a result, this design allows us to adequately capture both spatial and temporal features that are beneficial for detailed video understanding. Experimental results show that SF-LLaVA outperforms existing training-free methods on a wide range of video tasks. On some benchmarks, it achieves comparable or even better performance compared to state-of-the-art Video LLMs that are fine-tuned on video datasets. Code has been made available at: https://github.com/apple/ml-slowfast-llava.

discussion (0)

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

Forward citations

Cited by 24 Pith papers

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

  1. Molmo2: Open Weights and Data for Vision-Language Models with Video Understanding and Grounding

    cs.CV 2026-01 unverdicted novelty 8.0

    Molmo2 delivers state-of-the-art open-weight video VLMs with new grounding datasets and training methods that outperform prior open models and match or exceed some proprietary ones on pointing and tracking tasks.

  2. TrajTok: Learning Trajectory Tokens enables better Video Understanding

    cs.CV 2026-02 unverdicted novelty 7.0

    TrajTok learns adaptive trajectory tokens for videos through a unified end-to-end segmenter, improving understanding performance and efficiency over patch-based or external-pipeline tokenizers.

  3. WorldSense: Evaluating Real-world Omnimodal Understanding for Multimodal LLMs

    cs.CV 2025-02 unverdicted novelty 7.0

    WorldSense provides the first benchmark requiring synergistic audio-video-text understanding on 1,662 real-world videos and 3,172 QA pairs, where the best current multimodal LLM reaches only 65.1% accuracy.

  4. SkyVLaM: Multimodal Large Language Model for UAV Video Understanding in Remote Sensing

    cs.CV 2026-07 conditional novelty 6.0

    SkyVLaM introduces a temporal basis perceiver and adaptive dense selection to improve language-conditioned video segmentation in UAV scenes, and contributes the SkyVid dataset.

  5. Efficient Frame Selection for Long Videos at Test Time with Attention-Based MLLM Selectors

    cs.CV 2026-07 conditional novelty 6.0

    Attention maps from a small MLLM can serve as a training-free, query-conditioned frame selector, improving long-video QA accuracy under fixed frame budgets.

  6. Q-Fold: Query-Aware Focus-Context Spatio-Temporal Folding for Long Video Understanding

    cs.CV 2026-06 unverdicted novelty 6.0

    Q-Fold is a query-aware spatio-temporal folding technique that constructs heterogeneous focus-context inputs from long videos to improve Video-MLLM performance under fixed visual budgets.

  7. MemoryCard: Topic-Aware Multi-Modal Clue Compression for Long-Video Question Answering

    cs.CV 2026-06 unverdicted novelty 6.0

    MemoryCard organizes long videos into self-contained topic-aware Memory Cards that improve long-video QA accuracy by up to 21.8% relative under fixed visual-token budgets.

  8. V-LynX: Token Interface Alignment for Video+X LLMs

    cs.CV 2026-05 unverdicted novelty 6.0

    V-LynX integrates novel modalities into frozen Video LLMs by aligning to an internalized continuous token manifold using unpaired unimodal data and attention/statistical matching.

  9. WindowQuant: Mixed-Precision KV Cache Quantization based on Window-Level Similarity for VLMs Inference Optimization

    cs.CV 2026-05 unverdicted novelty 6.0

    WindowQuant performs window-adaptive mixed-precision KV cache quantization guided by similarity to the text prompt, with reordering to enable efficient inference in VLMs.

  10. One Token per Highly Selective Frame: Towards Extreme Compression for Long Video Understanding

    cs.CV 2026-04 unverdicted novelty 6.0

    XComp reaches extreme video compression (one token per selective frame) via learnable progressive token compression and question-conditioned frame selection, lifting LVBench accuracy from 42.9 percent to 46.2 percent ...

  11. TrajTok: Learning Trajectory Tokens enables better Video Understanding

    cs.CV 2026-02 conditional novelty 6.0

    TrajTok learns to tokenize video into object-trajectory tokens end-to-end, improving video CLIP, probing, and VLM performance over patch and token-merging baselines.

  12. LiveVLM: Efficient Online Video Understanding via Streaming-Oriented KV Cache and Retrieval

    cs.CV 2025-05 unverdicted novelty 6.0

    LiveVLM introduces VSB and PaR to compress and retrieve KV cache in streaming video LLMs, enabling LLaVA-OneVision to reach SOTA accuracy among training-free query-agnostic and training-based online models.

  13. LongVU: Spatiotemporal Adaptive Compression for Long Video-Language Understanding

    cs.CV 2024-10 unverdicted novelty 6.0

    LongVU adaptively compresses long video tokens using DINOv2-based frame deduplication, text-guided cross-modal selection, and temporal spatial reduction to improve video-language understanding in MLLMs with minimal de...

  14. LLaVA-Video: Video Instruction Tuning With Synthetic Data

    cs.CV 2024-10 unverdicted novelty 6.0

    LLaVA-Video-178K is a new synthetic video instruction dataset that, when combined with existing data to train LLaVA-Video, produces strong results on video understanding benchmarks.

  15. Mage-VL: An Efficient Codec-Native Streaming Multimodal Foundation Model

    cs.CV 2026-07 conditional novelty 5.0

    Codec-guided sparse patch selection plus a lightweight speak/silent gate yields a 4B streaming VLM that is competitive on static tasks, stronger on video/spatial benchmarks, and much cheaper at inference.

  16. ViCoStream: Streaming VideoLLMs Can Run Beyond 100 FPS with Stage-Wise Coordinated Inference

    cs.CV 2026-06 unverdicted novelty 5.0

    ViCoStream is a new coordinated pipeline framework for streaming VideoLLMs that achieves 134 FPS video throughput and less than 50 ms TTFT on A100 while keeping accuracy near full-history baselines.

  17. Linear Scaling Video VLMs for Long Video Understanding

    cs.CV 2026-05 unverdicted novelty 5.0

    StateKV is an inference-time technique that replaces quadratic self-attention prefill in video VLMs with a fixed-capacity importance-based recurrent state, keeping accuracy near full attention on long-video benchmarks...

  18. VidPrism: Heterogeneous Mixture of Experts for Image-to-Video Transfer

    cs.CV 2026-05 unverdicted novelty 5.0

    VidPrism introduces a heterogeneous temporal MoE with content-aware multi-rate sampling and bidirectional fusion for image-to-video transfer, claiming SOTA results on video benchmarks.

  19. EgoSelf: From Memory to Personalized Egocentric Assistant

    cs.CV 2026-04 unverdicted novelty 5.0

    EgoSelf uses graph-based memory of user interactions to derive personalized profiles and predict future behaviors for egocentric assistants.

  20. MASS: Motion-Aware Spatial-Temporal Grounding for Physics Reasoning and Comprehension in Vision-Language Models

    cs.CV 2025-11 unverdicted novelty 5.0

    MASS adds spatiotemporal motion signals and 3D grounding to VLMs and releases MASS-Bench, yielding physics-reasoning performance within 2% of Gemini-2.5-Flash after reinforcement fine-tuning.

  21. Video Parallel Scaling: Aggregating Diverse Frame Subsets for VideoLLMs

    cs.CV 2025-09 unverdicted novelty 5.0

    Video Parallel Scaling improves VideoLLM performance by aggregating outputs from parallel inferences on complementary disjoint frame subsets, effectively contracting the Chinchilla scaling law via uncorrelated visual ...

  22. NVILA: Efficient Frontier Visual Language Models

    cs.CV 2024-12 unverdicted novelty 5.0

    NVILA improves on VILA with a scale-then-compress visual token strategy and full-lifecycle efficiency optimizations, matching or exceeding leading VLMs on image and video benchmarks while reducing training cost 1.9-5....

  23. APB-V: Accelerating Long-Video Understanding via Sequence-Parallelism-aware Approximate Attention

    cs.CV 2026-01 conditional novelty 4.0

    SPAVA compresses each GPU's video context into essential key-values for sharing, making multi-GPU long-video prefilling 1.18x faster than APB and 12.72x faster than single-GPU FlashAttention with near-equal accuracy.

  24. VideoLLaMA 3: Frontier Multimodal Foundation Models for Image and Video Understanding

    cs.CV 2025-01 unverdicted novelty 4.0

    VideoLLaMA3 uses a vision-centric training paradigm and token-reduction design to reach competitive results on image and video benchmarks.