Temporal Video Understanding

Multimodal Large Language Models (MLLMs) are a major focus of recent AI research. However, most prior work focuses on static image understanding, while their ability to process sequential audio-video data remains underexplored. This gap highlights the need for a high-quality benchmark to systematically evaluate MLLM performance in a real-world setting. We introduce SONIC-O1, a comprehensive, fully human-verified benchmark of 60 hours (231 clips) spanning 13 real-world conversational domains with 4,958 annotations and demographic metadata. SONIC-O1 evaluates three capabilities: open-ended summarization, multiple-choice question (MCQ) answering, and temporal localization with supporting rationales (reasoning). Across closed- and open-source models, we find that the MCQ accuracy shows the smallest gap between model families, but the best closed-source model outperforms the best open-source model by 22.6% on temporal localization. We further observe accuracy gaps of up to 21.4% on temporal localization across demographic groups, indicating persistent disparities in model behaviour. SONIC-O1 provides an open evaluation suite for temporally grounded and demographically robust multimodal understanding. SONIC-O1 is publicly available for research: Project page (https://vectorinstitute.github.io/sonic-o1/), Dataset (https://huggingface.co/datasets/vector-institute/sonic-o1), GitHub (https://github.com/vectorinstitute/sonic-o1), Leaderboard (https://huggingface.co/spaces/vector-institute/sonic-o1-leaderboard).

Foundation models are used for many real-world applications involving language understanding and generation from temporally-ordered multimodal events. In this work, we study the ability of models to identify the most important sub-events in a video, which is a fundamental prerequisite for summarizing, reasoning over, or narrating multimodal events. Specifically, we focus on football games and evaluate models on their ability to distinguish between important and non-important sub-events in a game. To this end, we construct a new dataset by leveraging publicly available football highlight reels containing the most important moments of a game according to professional experts, without any additional annotation costs. Using our dataset, we compare several state-of-the-art multimodal models and show that they fail to reliably capture contextual importance. Analyses of models beyond standard evaluation metrics reveal their tendency to rely on a single dominant modality and their ineffectiveness in synthesizing necessary information from multiple sources. Our findings underline the importance of modular architectures that dynamically adapt to multimodal heterogeneity and the need for complementary training procedures that maximize cross-modal synergy.

Automatic identification of events and recurrent behavior analysis are critical for video surveillance. However, most existing content-based video retrieval benchmarks focus on scene-level similarity and do not evaluate the action discrimination required in surveillance. To address this gap, we introduce SOVABench (Surveillance Opposite Vehicle Actions Benchmark), a real-world retrieval benchmark built from surveillance footage and centered on vehicle-related actions. SOVABench defines two evaluation protocols (inter-pair and intra-pair) to assess cross-action discrimination and temporal direction understanding. Although action distinctions are generally intuitive for human observers, our experiments show that they remain challenging for state-of-the-art vision and multimodal models. Leveraging the visual reasoning and instruction-following capabilities of Multimodal Large Language Models (MLLMs), we present a training-free framework for producing interpretable embeddings from MLLM-generated descriptions for both images and videos. The framework achieves strong performance on SOVABench as well as on several spatial and counting benchmarks where contrastive Vision-Language Models often fail. The code, annotations, and instructions to construct the benchmark are publicly available.

With the rapid growth of video centered social media, the ability to anticipate risky events from visual data is a promising direction for ensuring public safety and preventing real world accidents. Prior work has extensively studied supervised video risk assessment across domains such as driving, protests, and natural disasters. However, many existing datasets provide models with access to the full video sequence, including the accident itself, which substantially reduces the difficulty of the task. To better reflect real world conditions, we introduce a new video understanding benchmark RiskCueBench in which videos are carefully annotated to identify a risk signal clip, defined as the earliest moment that indicates a potential safety concern. Experimental results reveal a significant gap in current systems ability to interpret evolving situations and anticipate future risky events from early visual signals, highlighting important challenges for deploying video risk prediction models in practice.

Multimodal large language models (MLLMs) have achieved impressive progress in vision-language reasoning, yet their ability to understand temporally unfolding narratives in videos remains underexplored. True narrative understanding requires grounding who is doing what, when, and where, maintaining coherent entity representations across dynamic visual and temporal contexts. We introduce NarrativeTrack, the first benchmark to evaluate narrative understanding in MLLMs through fine-grained entity-centric reasoning. Unlike existing benchmarks limited to short clips or coarse scene-level semantics, we decompose videos into constituent entities and examine their continuity via a Compositional Reasoning Progression (CRP), a structured evaluation framework that progressively increases narrative complexity across three dimensions: entity existence, entity changes, and entity ambiguity. CRP challenges models to advance from temporal persistence to contextual evolution and fine-grained perceptual reasoning. A fully automated entity-centric pipeline enables scalable extraction of temporally grounded entity representations, providing the foundation for CRP. Evaluations of state-of-the-art MLLMs reveal that models fail to robustly track entities across visual transitions and temporal dynamics, often hallucinating identity under context shifts. Open-source general-purpose MLLMs exhibit strong perceptual grounding but weak temporal coherence, while video-specific MLLMs capture temporal context yet hallucinate entities' contexts. These findings uncover a fundamental trade-off between perceptual grounding and temporal reasoning, indicating that narrative understanding emerges only from their integration. NarrativeTrack provides the first systematic framework to diagnose and advance temporally grounded narrative comprehension in MLLMs.

Open-Vocabulary Temporal Action Localization (OV-TAL) aims to recognize and localize instances of any desired action categories in videos without explicitly curating training data for all categories. Existing methods mostly recognize action categories at a single granularity, which degrades the recognition accuracy of both base and novel action categories. To address these issues, we propose a Multi-Grained Category-Aware Network (MGCA-Net) comprising a localizer, an action presence predictor, a conventional classifier, and a coarse-to-fine classifier. Specifically, the localizer localizes category-agnostic action proposals. For these action proposals, the action presence predictor estimates the probability that they belong to an action instance. At the same time, the conventional classifier predicts the probability of each action proposal over base action categories at the snippet granularity. Novel action categories are recognized by the coarse-to-fine classifier, which first identifies action presence at the video granularity. Finally, it assigns each action proposal to one category from the coarse categories at the proposal granularity. Through coarse-to-fine category awareness for novel actions and the conventional classifier's awareness of base actions, multi-grained category awareness is achieved, effectively enhancing localization performance. Comprehensive evaluations on the THUMOS'14 and ActivityNet-1.3 benchmarks demonstrate that our method achieves state-of-the-art performance. Furthermore, our MGCA-Net achieves state-of-the-art results under the Zero-Shot Temporal Action Localization setting.

Recent advances in video large language models (VideoLLMs) call for new evaluation protocols and benchmarks for video-language understanding. Inspired by the needle in a haystack test widely used by LLMs, we introduce a novel task of Needle in a Montage (NeMo), which is designed to assess the temporal understanding capabilities of advanced VideoLLMs. Specifically, the proposed task focuses on two fundamental abilities critical for temporal understanding, i.e., retrieval-style long-context recall and temporal grounding. To generate video question answering data for our task, we develop a scalable automated data generation pipeline that facilitates high-quality data synthesis. Built upon the proposed pipeline, we present NeMoBench, a video-language benchmark centered on our task. Specifically, our full set of NeMoBench features 31,378 automatically generated question-answer (QA) pairs from 13,486 videos with various durations ranging from seconds to hours. Experiments demonstrate that our pipeline can reliably and automatically generate high-quality evaluation data, enabling NeMoBench to be continuously updated with the latest videos. We evaluate 20 state-of-the-art models on our benchmark, providing extensive results and key insights into their capabilities and limitations. Our project page is available at: https://lavi-lab.github.io/NeMoBench.

Vision-language models (VLMs) can ingest only a limited number of video frames, making frame selection a practical necessity. But do current Video QA benchmarks genuinely require temporal frame selection, or can most questions be answered regardless of which frames are shown? We introduce Frame Selection Sensitivity (FSS), a per-sample diagnostic that measures how much VLM accuracy changes when the most relevant frames are replaced with the least relevant ones. Across six benchmarks and eight VLMs, we find that a large majority of samples are frame-agnostic: only a minority are genuinely sensitive to frame choice. Combining FSS with a Language Independence Score (LIS) reveals that merely 5.5--31% of samples are Temporally Sensitive. We construct TempCore, compact evaluation subsets that isolate these temporal samples from existing benchmarks, and will release code and per-sample annotations upon publication.

Video Large Language Models (VLLMs) excel in video understanding, but their excessive visual tokens pose a significant computational challenge for real-world applications. Current methods aim to enhance inference efficiency by visual token pruning. However, they do not consider the dynamic characteristics and temporal dependencies of video frames, as they perceive video understanding as a multi-frame task. To address these challenges, we propose MMG-Vid, a novel training-free visual token pruning framework that removes redundancy by Maximizing Marginal Gains at both segment-level and token-level. Specifically, we first divide the video into segments based on frame similarity, and then dynamically allocate the token budget for each segment to maximize the marginal gain of each segment. Subsequently, we propose a temporal-guided DPC algorithm that jointly models inter-frame uniqueness and intra-frame diversity, thereby maximizing the marginal gain of each token. By combining both stages, MMG-Vid can maximize the utilization of the limited token budget, significantly improving efficiency while maintaining strong performance. Extensive experiments demonstrate that MMG-Vid can maintain over 99.5% of the original performance, while effectively reducing 75% visual tokens and accelerating the prefilling stage by 3.9x on LLaVA-OneVision-7B. Code is available at https://github.com/ylnxxts/MMG-Vid.

The rapid proliferation of video in applications such as autonomous driving, surveillance, and sports analytics necessitates robust methods for dynamic scene understanding. Despite advances in static scene graph generation and early attempts at video scene graph generation, previous methods often suffer from fragmented representations, failing to capture fine-grained spatial details and long-range temporal dependencies simultaneously. To address these limitations, we introduce the Temporal Hierarchical Cyclic Scene Graph (THYME) approach, which synergistically integrates hierarchical feature aggregation with cyclic temporal refinement to address these limitations. In particular, THYME effectively models multi-scale spatial context and enforces temporal consistency across frames, yielding more accurate and coherent scene graphs. In addition, we present AeroEye-v1.0, a novel aerial video dataset enriched with five types of interactivity that overcome the constraints of existing datasets and provide a comprehensive benchmark for dynamic scene graph generation. Empirically, extensive experiments on ASPIRe and AeroEye-v1.0 demonstrate that the proposed THYME approach outperforms state-of-the-art methods, offering improved scene understanding in ground-view and aerial scenarios.

Understanding causal event relationships and achieving fine-grained temporal grounding in videos remain challenging for vision-language models (VLMs). We propose TEMPURA (Temporal Event Masked Prediction and Understanding for Reasoning in Action), a two-stage training framework that enhances the video temporal understanding of VLMs. Inspired by infilling techniques in language modeling, TEMPURA first performs masked event prediction, learning to reconstruct missing events and generate step-by-step causal explanations from dense event annotations. It then learns video segmentation and dense captioning, decomposing videos into non-overlapping events with detailed, timestamp-aligned descriptions. We train TEMPURA on VER, our large-scale dataset of 500K videos annotated with temporally aligned event descriptions and structured reasoning steps. Experiments on video temporal grounding and highlight detection benchmarks show that TEMPURA substantially improves strong base VLMs across model families and scales, confirming that combining event-level reasoning with fine-grained temporal segmentation is an effective recipe for video temporal understanding.