COMET: Contrastive Motion-Enhanced Temporal Reasoning for Video Multimodal Large Language Models

• URL: http://arxiv.org/abs/2608.21030v1
• Date: Fri, 21 Aug 2026 12:28:36 GMT
• Title: COMET: Contrastive Motion-Enhanced Temporal Reasoning for Video Multimodal Large Language Models
• Abstract summary: Video multimodal large language models have advanced significantly, yet fine-grained motion-temporal understanding remains fragile.<n>We propose COMET, a temporally grounded framework that systematically strengthens video MLLMs through explicit temporal representation.<n>For optimization, COMET combines temporal prior distillation with a forward-reverse TC-GRPO stage that turns temporal order into a direct learning signal.
• Score: 15.39037309359101
• License: http://arxiv.org/licenses/nonexclusive-distrib/1.0/
• Abstract: Video multimodal large language models have advanced significantly, yet fine-grained motion-temporal understanding remains fragile. The core bottleneck is not only sparse frame sampling, but also the lack of a complete temporal modeling pipeline for explicitly representing frame-to-frame change, enabling appearance-motion interaction, and optimizing temporal direction sensitivity. We propose COMET, a temporally grounded framework that systematically strengthens video MLLMs through explicit temporal representation, appearance-motion fusion, and direction-aware optimization. Architecturally, COMET introduces a temporal motion branch built on Taylor frame differences and injects its motion evidence into the appearance stream via temporal attention bias-enhanced cross-attention. For optimization, COMET combines temporal prior distillation with a forward-reverse TC-GRPO stage that turns temporal order into a direct learning signal and strengthens the model's use of directional motion patterns encoded by the temporal motion branch. The method achieves consistent overall improvements with a pronounced motion-temporal bias: on Qwen3-VL-8B, action-centric tasks (STAR, SSv2) improve by 4.9% on average, temporal reasoning tasks (NExT-QA, CLEVRER, LLaVA-178K) by 2.1% over BL-GRPO, while static perception tasks (PerceptionTest) remain on par. The same gain pattern also transfers to InternVL2.5-8B, indicating that COMET generalizes across model families.

Related papers

• StreamPI: Streaming Multimodal Temporal Modeling for Vision-Language-Action Models [84.17128030384099]
  StreamPI is a streaming multimodal temporal modeling framework.<n>It equips single-frame VLA with temporal reasoning capability without introducing any additional parameters.<n> Experiments on real-robot tasks spanning memory-dependent and precise perception scenarios, as well as the simulation benchmark LIBERO, demonstrate that StreamPI outperforms pi0.5 across diverse tasks.
  arXiv  Detail & Related papers  (2026-08-26T17:33:19Z)
• Can Video Diffusion Models Predict Past Frames? Bidirectional Cycle Consistency for Reversible Interpolation [25.677744104220853]
  Video frame aims to synthesize realistic intermediate frames between given endpoints while adhering to specific motion semantics.<n>We propose a novel bidirectional framework that enforces symmetry between forward and backward generation trajectories.<n>Our method achieves state-of-the-art performance in imaging quality, motion smoothness, and dynamic control on both 37-frame and 73-frame tasks.
  arXiv  Detail & Related papers  (2026-04-02T06:58:46Z)
• Towards Arbitrary Motion Completing via Hierarchical Continuous Representation [64.6525112550758]
  We propose a novel parametric activation-induced hierarchical implicit representation framework, called NAME, based on Implicit Representations (INRs)<n>Our method introduces a hierarchical temporal encoding mechanism that extracts features from motion sequences at multiple temporal scales, enabling effective capture of intricate temporal patterns.
  arXiv  Detail & Related papers  (2025-12-24T14:07:04Z)
• Bidirectional Feature-aligned Motion Transformation for Efficient Dynamic Point Cloud Compression [97.66080040613726]
  We propose a Bidirectional Feature-aligned Motion Transformation (Bi-FMT) framework that implicitly models motion in the feature space.<n>Bi-FMT aligns features across both past and future frames to produce temporally consistent latent representations.<n>We show Bi-FMT surpasses D-DPCC and AdaDPCC in both compression efficiency and runtime.
  arXiv  Detail & Related papers  (2025-09-18T03:51:06Z)
• FlowMo: Variance-Based Flow Guidance for Coherent Motion in Video Generation [51.110607281391154]
  FlowMo is a training-free guidance method for enhancing motion coherence in text-to-video models.<n>It estimates motion coherence by measuring the patch-wise variance across the temporal dimension and guides the model to reduce this variance dynamically during sampling.
  arXiv  Detail & Related papers  (2025-06-01T19:55:33Z)
• Motion-Aware Generative Frame Interpolation [23.380470636851022]
  Flow-based frame methods ensure motion stability through estimated intermediate flow but often introduce severe artifacts in complex motion regions.<n>Recent generative approaches, boosted by large-scale pre-trained video generation models, show promise in handling intricate scenes.<n>We propose Motion-aware Generative frame (MoG) that synergizes intermediate flow guidance with generative capacities to enhance fidelity.
  arXiv  Detail & Related papers  (2025-01-07T11:03:43Z)
• Live2Diff: Live Stream Translation via Uni-directional Attention in Video Diffusion Models [64.2445487645478]
  Large Language Models have shown remarkable efficacy in generating streaming data such as text and audio. We present Live2Diff, the first attempt at designing a video diffusion model with uni-directional temporal attention, specifically targeting live streaming video translation.
  arXiv  Detail & Related papers  (2024-07-11T17:34:51Z)
• Generalizable Implicit Motion Modeling for Video Frame Interpolation [51.966062283735596]
  Motion is critical in flow-based Video Frame Interpolation (VFI)<n>We introduce General Implicit Motion Modeling (IMM), a novel and effective approach to motion modeling VFI.<n>Our GIMM can be easily integrated with existing flow-based VFI works by supplying accurately modeled motion.
  arXiv  Detail & Related papers  (2024-07-11T17:13:15Z)
• TSI: Temporal Saliency Integration for Video Action Recognition [32.18535820790586]
  We propose a Temporal Saliency Integration (TSI) block, which mainly contains a Salient Motion Excitation (SME) module and a Cross-scale Temporal Integration (CTI) module. SME aims to highlight the motion-sensitive area through local-global motion modeling. CTI is designed to perform multi-scale temporal modeling through a group of separate 1D convolutions respectively.
  arXiv  Detail & Related papers  (2021-06-02T11:43:49Z)
• Learning Self-Similarity in Space and Time as Generalized Motion for Action Recognition [42.175450800733785]
  We propose a rich motion representation based on video self-similarity (STSS) We leverage the whole volume of STSSS and let our model learn to extract an effective motion representation from it. The proposed neural block, dubbed SELFY, can be easily inserted into neural architectures and trained end-to-end without additional supervision.
  arXiv  Detail & Related papers  (2021-02-14T07:32:55Z)

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.