OTT-Vid: Optimal Transport Temporal Token Compression for Video Large Language Models

• URL: http://arxiv.org/abs/2605.11803v1
• Date: Tue, 12 May 2026 08:58:49 GMT
• Title: OTT-Vid: Optimal Transport Temporal Token Compression for Video Large Language Models
• Abstract summary: OTT-Vid is a transport-derived allocation framework for temporal token compression.<n>We show that OTT-Vid preserves 95.8% of VQA and 73.9% of VTG performance while retaining only 10% of tokens.
• Score: 39.853488828881986
• License: http://creativecommons.org/licenses/by/4.0/
• Abstract: As Video Large Language Models (Video-LLMs) scale to longer and more complex videos, their inference cost grows rapidly due to the large volume of visual tokens accumulated across frames. Training-free token compression has emerged as a practical solution to this bottleneck. However, existing temporal compression methods rely primarily on cross-frame token similarity or segmentation heuristics, overlooking each token's semantic role within its frame and failing to adapt compression strength to the compressibility of each frame pair. In this work, we propose OTT-Vid, a transport-derived allocation framework for temporal token compression. Our approach consists of two stages: spatial pruning identifies representative content within each frame, and optimal transport (OT) is then solved between neighboring frames to estimate temporal compressibility. We formulate this OT with non-uniform token mass, which protects semantically important tokens from aggressive compression, and a locality-aware cost that captures both feature and spatial disparities. The resulting transport plan jointly balances token importance and matching cost, while its total cost defines the transport difficulty of each frame pair, which we use to allocate compression budgets dynamically. Experiments on six benchmarks spanning video question answering and temporal grounding show that OTT-Vid preserves 95.8% of VQA and 73.9% of VTG performance while retaining only 10% of tokens, consistently outperforming existing state-of-the-art training-free compression methods.

Related papers

• Aggregating Visual Information with Optimal Transport for VideoLM Token Compression [55.867497983465476]
  Video language models process videos as dense visual-token sequences with substantial representational redundancy.<n>These sequences are therefore essential for reducing the visual-token burden on language-model decoding.<n>We introduce Aggregating Visual Information with Optimal Transport (AVIOT), which casts video token compression as transporting a dense empirical measure of frame observations.
  arXiv  Detail & Related papers  (2026-08-20T18:05:42Z)
• STAC: Selective Spatiotemporal Aggregation and Compression for Video Reasoning Segmentation [60.597357847359255]
  Video segmentation demands pixel-accurate object tracking across hundreds of frames under complex natural language queries.<n>Existing methods address this through token compression, yet typically operate on features lacking temporal context.<n>State-space models resolve this constraint, as their linear recurrence selectively conditions each token on temporal context at $mathcalO(T)$ cost compression to reduce.
  arXiv  Detail & Related papers  (2026-07-03T03:28:25Z)
• InfoMerge: Information-aware Token Compression for Efficient Video Large Language Models [24.36425066833425]
  Video Large Language Models (Video-LLMs) achieve strong performance in video understanding, but their excessive visual tokens bring substantial computational overhead.<n>We propose InfoMerge, a training-free visual token compression method that improves token utilization through robust redundancy estimation and content-aware budget allocation.<n>On LLaVA-OneVision-7B, InfoMerge retains 98.8% of the original average performance while reducing 85% of visual tokens and achieving a 4.24-fold speedup in the prefill stage.
  arXiv  Detail & Related papers  (2026-06-01T12:24:24Z)
• Small Vision-Language Models are Smart Compressors for Long Video Understanding [73.65465038390771]
  Long-form video understanding relies on intent-driven efficiency rather than greedily padded context.<n>We propose Tempo, an efficient query-aware framework compressing long videos for downstream understanding.<n>Tests show Tempo compresses hour-long videos substantially below theoretical limits, proving true long-form video understanding relies on intent-driven efficiency.
  arXiv  Detail & Related papers  (2026-04-09T11:40:25Z)
• Unified Spatiotemporal Token Compression for Video-LLMs at Ultra-Low Retention [23.015486635502437]
  Video language models (Video-LLMs) face high computational costs due to large volumes of visual tokens.<n>We propose a unified selection mechanism that weights semantic similarity to globally select tokens.<n>Unselected tokens are merged via clustering and refilled, preserving information integrity.<n>Our unifiedtemporal token compression strategy establishes the state-of-theart in video understanding under ultra-low token retention.
  arXiv  Detail & Related papers  (2026-03-23T13:15:22Z)
• Prune Redundancy, Preserve Essence: Vision Token Compression in VLMs via Synergistic Importance-Diversity [54.95089105944234]
  Vision-language models (VLMs) face significant computational inefficiencies caused by excessive generation of visual tokens.<n>We propose PruneSID, a training-free Synergistic Importance-Diversity approach featuring a two-stage pipeline.<n>We demonstrate state-of-the-art performance, achieving 96.3% accuracy on LLaVA-1.5 with only 11.1% token retention, and 92.8% accuracy at extreme compression rates (5.6%) on LLaVA-NeXT.
  arXiv  Detail & Related papers  (2026-03-10T10:31:58Z)
• Keeping the Evidence Chain: Semantic Evidence Allocation for Training-Free Token Pruning in Video Temporal Grounding [16.537148896706036]
  Video Temporal Grounding (VTG) localizes the temporal boundaries of a query-relevant moment in long, untrimmed videos.<n>We propose SemVID, a training-free pruning framework that constructs a compact yet coherent token subset with semantic roles.
  arXiv  Detail & Related papers  (2026-03-05T20:25:32Z)
• StreamingTOM: Streaming Token Compression for Efficient Video Understanding [6.9203477336374775]
  Existing approaches only regulate post-LLM kv-cache, leaving costly pre-LLM prefill unchanged.<n>We introduce StreamingTOM, a training-free, plug-and-play two-stage framework that addresses both pre-LLM and post-LLM bottlenecks with predictable latency.<n> Experiments demonstrate our method achieves $15.7times$ kv-cache compression, $1.2times$ lower peak memory and $2times$ faster TTFT compared to prior SOTA.
  arXiv  Detail & Related papers  (2025-10-21T03:39:41Z)
• PVC: Progressive Visual Token Compression for Unified Image and Video Processing in Large Vision-Language Models [64.9366388601049]
  Visual token compression is leveraged to reduce the considerable token length of visual inputs.<n>We introduce a unified token compression strategy called Progressive Visual Token Compression.<n>Our model achieves state-of-the-art performance across various video understanding benchmarks.
  arXiv  Detail & Related papers  (2024-12-12T18:59:40Z)
• CAIT: Triple-Win Compression towards High Accuracy, Fast Inference, and Favorable Transferability For ViTs [89.79139531731637]
  Vision Transformers (ViTs) have emerged as state-of-the-art models for various vision tasks.<n>We propose a joint underlinecompression method for ViTs that achieves a harmonious blend of high underlineaccuracy, fast underlineinference speed, and favorable underlinetransferability to downstream tasks.
  arXiv  Detail & Related papers  (2023-09-27T16:12:07Z)

This list is automatically generated from the titles and abstracts of the papers in this site.

This site does not guarantee the quality of this site (including all information) and is not responsible for any consequences.