mlsanityData to 14 Oct 2026

VideoChat-R1.5: Visual Test-Time Scaling to Reinforce Multimodal Reasoning by Iterative Perception

arXiv 2509.21100 ↗First posted 25 Sep 2025cs.CVcited by 65

On the map: LLM reasoning → Multimodal reasoning

also in 1 more
Abstract (arXiv)

Inducing reasoning in multimodal large language models (MLLMs) is critical for achieving human-level perception and understanding. Existing methods mainly leverage LLM reasoning to analyze parsed visuals, often limited by static perception stages. This paper introduces Visual Test-Time Scaling (VTTS), a novel approach to enhance MLLMs' reasoning via iterative perception during inference. VTTS mimics humans' hierarchical attention by progressively refining focus on high-confidence spatio-temporal regions, guided by updated textual predictions. Specifically, VTTS employs an Iterative Perception (ITP) mechanism, incorporating reinforcement learning with spatio-temporal supervision to optimize reasoning. To support this paradigm, we also present VTTS-80K, a dataset tailored for iterative perception. These designs allows a MLLM to enhance its performance by increasing its perceptual compute. Extensive experiments validate VTTS's effectiveness and generalization across diverse tasks and benchmarks. Our newly introduced Videochat-R1.5 model has achieved remarkable improvements, with an average increase of over 5%, compared to robust baselines such as Qwen2.5VL-3B and -7B, across more than 15 benchmarks that encompass video conversation, video reasoning, and spatio-temporal perception.

Key facts. Cited by 65 papers in the map (data through 14 Oct 2026). Placed in Multimodal reasoning, one of 17 directions in LLM reasoning. Builds on 64 of its 80 references that are in the map. First posted on arXiv 25 Sep 2025.

Builds on

64 of 80 references.

Closest ancestors · 15

Same direction, newest first

  1. 2025VL-Rethinker: Incentivizing Self-Reflection of Vision-Language Models with Reinforcement Learning10 Apr
  2. 2025VideoChat-R1: Enhancing Spatio-Temporal Perception via Reinforcement Fine-Tuning9 Apr
  3. 2025Video-R1: Reinforcing Video Reasoning in MLLMs27 Mar
  4. 2025R1-Onevision: Advancing Generalized Multimodal Reasoning through Cross-Modal Formalization13 Mar
  5. 2025VisualPRM: An Effective Process Reward Model for Multimodal Reasoning13 Mar
  6. 2025Seg-Zero: Reasoning-Chain Guided Segmentation via Cognitive Reinforcement9 Mar
  7. 2025R1-Zero's "Aha Moment" in Visual Reasoning on a 2B Non-SFT Model7 Mar
  8. 2024Expanding Performance Boundaries of Open-Source Multimodal Models with Model, Data, and Test-Time Scaling6 Dec
  9. 2024Insight-V: Exploring Long-Chain Visual Reasoning with Multimodal Large Language Models21 Nov
  10. 2024LLaVA-CoT: Let Vision Language Models Reason Step-by-Step15 Nov
  11. 2024ReXTime: A Benchmark Suite for Reasoning-Across-Time in Videos27 Jun
  12. 2024VideoTree: Adaptive Tree-based Video Representation for LLM Reasoning on Long Videos29 May
3 more
  1. 2024Visual CoT: Advancing Multi-Modal Language Models with a Comprehensive Dataset and Benchmark for Chain-of-Thought Reasoning25 Mar
  2. 2023Perception Test: A Diagnostic Benchmark for Multimodal Video Models23 May
  3. 2022Learn to Explain: Multimodal Reasoning via Thought Chains for Science Question Answering20 Sep

Other references · 49

  1. 2023GPT-4 Technical Reportcited by 18,396
  2. 2023Visual Instruction Tuningcited by 7,869
  3. 2023BLIP-2: Bootstrapping Language-Image Pre-training with Frozen Image Encoders and Large Language Modelscited by 6,137
  4. 2025Qwen2.5-VL Technical Reportcited by 5,587
  5. 2024Qwen2-VL: Enhancing Vision-Language Model's Perception of the World at Any Resolutioncited by 4,234
  6. 2023Grounding DINO: Marrying DINO with Grounded Pre-Training for Open-Set Object Detectioncited by 3,332
  7. 2024LLaVA-OneVision: Easy Visual Task Transfercited by 2,719
  8. 2023InternVL: Scaling up Vision Foundation Models and Aligning for Generic Visual-Linguistic Taskscited by 2,565
  9. 2023Video-LLaMA: An Instruction-tuned Audio-Visual Language Model for Video Understandingcited by 1,471
  10. 2024Video-MME: The First-Ever Comprehensive Evaluation Benchmark of Multi-modal LLMs in Video Analysiscited by 1,413
  11. 2023MVBench: A Comprehensive Multi-modal Video Understanding Benchmarkcited by 1,109
  12. 2016Modeling Context in Referring Expressionscited by 1,063
37 more
  1. 2023mPLUG-Owl: Modularization Empowers Large Language Models with Multimodalitycited by 1,040
  2. 2015Generation and Comprehension of Unambiguous Object Descriptionscited by 904
  3. 2023VideoChat: Chat-Centric Video Understandingcited by 857
  4. 2021MDETR -- Modulated Detection for End-to-End Multi-Modal Understandingcited by 819
  5. 2024VideoLLaMA 2: Advancing Spatial-Temporal Modeling and Audio Understanding in Video-LLMscited by 771
  6. 2024LongVideoBench: A Benchmark for Long-context Interleaved Video-Language Understandingcited by 665
  7. 2024Thinking in Space: How Multimodal Large Language Models See, Remember, and Recall Spacescited by 611
  8. 2017TALL: Temporal Activity Localization via Language Querycited by 584
  9. 2016Fully-Convolutional Siamese Networks for Object Trackingcited by 568
  10. 2025VideoLLaMA 3: Frontier Multimodal Foundation Models for Image and Video Understandingcited by 518
  11. 2023V*: Guided Visual Search as a Core Mechanism in Multimodal LLMscited by 482
  12. 2023OCRBench: On the Hidden Mystery of OCR in Large Multimodal Modelscited by 476
  13. 2018GOT-10k: A Large High-Diversity Benchmark for Generic Object Tracking in the Wildcited by 475
  14. 2022InternVideo: General Video Foundation Models via Generative and Discriminative Learningcited by 447
  15. 2024LVBench: An Extreme Long Video Understanding Benchmarkcited by 410
  16. 2023TimeChat: A Time-sensitive Multimodal Large Language Model for Long Video Understandingcited by 406
  17. 2024InternVideo2: Scaling Foundation Models for Multimodal Video Understandingcited by 317
  18. 2025R1-VL: Learning to Reason with Multimodal Large Language Models via Step-wise Group Relative Policy Optimizationcited by 301
  19. 2023Alphazero-like Tree-Search can Guide Large Language Model Decoding and Trainingcited by 295
  20. 2025Video-MMMU: Evaluating Knowledge Acquisition from Multi-Discipline Professional Videoscited by 236
  21. 2024VideoChat-Flash: Hierarchical Compression for Long-Context Video Modelingcited by 220
  22. 2025InternVideo2.5: Empowering Video MLLMs with Long and Rich Context Modelingcited by 211
  23. 2023G-LLaVA: Solving Geometric Problem with Multi-Modal Large Language Modelcited by 210
  24. 2023Self-Chained Image-Language Model for Video Localization and Question Answeringcited by 203
  25. 2023Self-Evaluation Guided Beam Search for Reasoningcited by 196
  26. 2023UniVTG: Towards Unified Video-Language Temporal Groundingcited by 190
  27. 2023A Simple LLM Framework for Long-Range Video Question-Answeringcited by 178
  28. 2024Streaming Long Video Understanding with Large Language Modelscited by 167
  29. 2023Can I Trust Your Answer? Visually Grounded Video Question Answeringcited by 135
  30. 2024BoNBoN Alignment for Large Language Models and the Sweetness of Best-of-n Samplingcited by 121
  31. 2023The All-Seeing Project: Towards Panoptic Visual Recognition and Understanding of the Open Worldcited by 116
  32. 2024TimeSuite: Improving MLLMs for Long Video Understanding via Grounded Tuningcited by 106
  33. 2025R1-Omni: Explainable Omni-Multimodal Emotion Recognition with Reinforcement Learningcited by 100
  34. 2024The All-Seeing Project V2: Towards General Relation Comprehension of the Open Worldcited by 90
  35. 2022An Empirical Study of End-to-End Video-Language Transformers with Masked Visual Modelingcited by 74
  36. 2024Language Repository for Long Video Understandingcited by 55
  37. 2024Task Preference Optimization: Improving Multimodal Large Language Models with Vision Task Alignmentcited by 29

Built on by

Cited by year (2026 to 14 Oct)

Citing papers per year: 2025: 9; 2026: 56.

65 papers.

Sort:

By area:

  1. 2026Video-MME-v2: Towards the Next Stage in Benchmarks for Comprehensive Video Understandingcited by 55
  2. 2026Video-OPD: Efficient Post-Training of Multimodal Large Language Models for Temporal Video Grounding via On-Policy Distillationcited by 36
  3. 2025Skyra: AI-Generated Video Detection via Grounded Artifact Reasoningcited by 25
  4. 2026Video-o3: Native Interleaved Clue Seeking for Long Video Multi-Hop Reasoningcited by 24
  5. 2026Video Streaming Thinking: VideoLLMs Can Watch and Think Simultaneouslycited by 22
  6. 2026VideoAuto-R1: Video Auto Reasoning via Thinking Once, Answering Twicecited by 21
  7. 2025DiffThinker: Towards Generative Multimodal Reasoning with Diffusion Modelscited by 19
  8. 2025REVISOR: Beyond Textual Reflection, Towards Multimodal Introspective Reasoning in Long-Form Video Understandingcited by 16
  9. 2026When and How Much to Imagine: Adaptive Test-Time Scaling with World Models for Visual Spatial Reasoningcited by 14
  10. 2025Thinking With Bounding Boxes: Enhancing Spatio-Temporal Video Grounding via Reinforcement Fine-Tuningcited by 12
  11. 2026VISD: Enhancing Video Reasoning via Structured Self-Distillationcited by 10
  12. 2026VideoChat3: Fully Open Video MLLM for Efficient and Generalist Video Understandingcited by 10
38 more
  1. 2025TempR1: Improving Temporal Understanding of MLLMs via Temporal-Aware Multi-Task Reinforcement Learningcited by 8
  2. 2025MedGRPO: Multi-Task Reinforcement Learning for Heterogeneous Medical Video Understandingcited by 8
  3. 2026VideoTemp-o3: Harmonizing Temporal Grounding and Video Understanding in Agentic Thinking-with-Videoscited by 7
  4. 2026VideoZeroBench: Probing the Limits of Video MLLMs with Spatio-Temporal Evidence Verificationcited by 7
  5. 2025SAGE: Training Smart Any-Horizon Agents for Long Video Reasoning with Reinforcement Learningcited by 6
  6. 2026EndoCoT: Scaling Endogenous Chain-of-Thought Reasoning in Diffusion Modelscited by 6
  7. 2026Video Understanding: From Geometry and Semantics to Unified Modelscited by 5
  8. 2026MM-ReCoder: Advancing Chart-to-Code Generation with Reinforcement Learning and Self-Correctioncited by 5
  9. 2026Moment-Video: Diagnosing Temporal Fidelity of Video MLLMs on Momentary Visual Eventscited by 5
  10. 2026VisionCoach: Reinforcing Grounded Video Reasoning via Visual-Perception Promptingcited by 4
  11. 2026EvoGround: Self-Evolving Video Agents for Video Temporal Groundingcited by 4
  12. 2026Towards One-to-Many Temporal Groundingcited by 4
  13. 2026Watch, Remember, Reason: Human-View Video Understanding with MLLMscited by 4
  14. 2026Building a Precise Video Language with Human-AI Oversightcited by 3
  15. 2026EgoPro-Bench: Benchmarking Personalized Proactive Interaction in Egocentric Video Streamscited by 3
  16. 2026Temporal-Aware Reasoning Optimization for Video Temporal Groundingcited by 3
  17. 2025Thinking with Drafts: Speculative Temporal Reasoning for Efficient Long Video Understandingcited by 2
  18. 2026Graph-to-Frame RAG: Visual-Space Knowledge Fusion for Training-Free and Auditable Video Reasoningcited by 2
  19. 2026Reinforce to Learn, Elect to Reason: A Dual Paradigm for Video Reasoningcited by 2
  20. 2026Audio-Visual Intelligence in Large Foundation Modelscited by 2
  21. 2026CineCap: Structured Reasoning with Spatio-Temporal Anchors for Cinematographic Video Captioningcited by 2
  22. 2026MVP: Enhancing Video Large Language Models via Self-supervised Masked Video Predictioncited by 1
  23. 2026Perception Without Engagement: Dissecting the Causal Discovery Deficit in LMMscited by 1
  24. 2026GRASP: Learning to Ground Social Reasoning in Multi-Person Non-Verbal Interactionscited by 1
  25. 2026VideoOdyssey: A Benchmark for Ultra-Long-Context and Omni-Modal Video Understandingcited by 1
  26. 2026EG-VQA: Benchmarking Verifiable Video Question Answering with Grounded Temporal Evidencecited by 1
  27. 2026VideoSearch-R1: Iterative Video Retrieval and Reasoning via Soft Query Refinementcited by 1
  28. 2026Be Indiscrete: The Benefits of Learning Continuous Spine Degeneration Severity Scorescited by 1
  29. 2026TAU-Bench: From Anomaly Instance Tracking to Fine-Grained Video Anomaly Understandingcited by 1
  30. 2026FluxLite: Inference-Time Proposal Control for Discrete Diffusion Modelscited by 1
  31. 2025Decoupling Perception from Reasoning for Hallucination-Resistant Video Understandingcited by 0
  32. 2026CASHEW: Stabilizing Multimodal Reasoning via Iterative Trajectory Aggregationcited by 0
  33. 2026TTA-Vid: Generalized Test-Time Adaptation for Video Reasoningcited by 0
  34. 2026TOC-Bench: A Temporal Object Consistency Benchmark for Video Large Language Modelscited by 0
  35. 2026MMVIAD: Multi-view Multi-task Video Understanding for Industrial Anomaly Detectioncited by 0
  36. 2026Test-Time Scaling in Multimodal Foundation Models: A Comprehensive Survey of Generation and Reasoningcited by 0
  37. 2026Counterfactual Reasoning for Fine-Grained Evidence Disentanglement in VideoQAcited by 0
  38. 2026VideoLatent: Video-Language Learning via Latent Self-Forcingcited by 0

50 most cited of 65 All 65 (JSON)

Similar, not cited

Close in wording, no citation either way

  1. 2025VTPerception-R1: Enhancing Multimodal Reasoning via Explicit Visual and Textual Perceptual Grounding
  2. 2025Unleashing Perception-Time Scaling to Multimodal Reasoning Models
  3. 2025Limits and Gains of Test-Time Scaling in Vision-Language Reasoning
  4. 2025Perception-R1: Advancing Multimodal Reasoning Capabilities of MLLMs via Visual Perception Reward
  5. 2026See It, Say It, Sorted: An Iterative Training-Free Framework for Visually-Grounded Multimodal Reasoning in LVLMs
  6. 2025VKnowU: Evaluating Visual Knowledge Understanding in Multimodal LLMs
  7. 2026Do MLLMs Really See It: Reinforcing Visual Attention in Multimodal LLMs
  8. 2026Faithful-MR1: Faithful Multimodal Reasoning via Anchoring and Reinforcing Visual Attention