The advancements in large language models (LLMs) have propelled the improvement of video understanding tasks by incorporating LLMs with visual models. However, most existing LLM-based models (e.g., VideoLLaMA, VideoChat) are constrained to processing short-duration videos. Recent attempts to understand long-term videos by extracting and compressing visual features into a fixed memory size. Nevertheless, those methods leverage only visual modality to merge video tokens and overlook the correlation between visual and textual queries, leading to difficulties in effectively handling complex question-answering tasks. To address the challenges of long videos and complex prompts, we propose AdaCM$^2$, which, for the first time, introduces an adaptive cross-modality memory reduction approach to video-text alignment in an auto-regressive manner on video streams. Our extensive experiments on various video understanding tasks, such as video captioning, video question answering, and video classification, demonstrate that AdaCM$^2$ achieves state-of-the-art performance across multiple datasets while significantly reducing memory usage. Notably, it achieves a 4.5% improvement across multiple tasks in the LVU dataset with a GPU memory consumption reduction of up to 65%.
Recent advances in multimodal Large Language Models (LLMs) have shown great success in understanding multi-modal contents. For video understanding tasks, training-based video LLMs are difficult to build due to the scarcity of high-quality, curated video-text paired data. In contrast, paired image-text data are much easier to obtain, and there is substantial similarity between images and videos. Consequently, extending image LLMs for video understanding tasks presents an appealing alternative. Developing effective strategies for compressing visual tokens from multiple frames is a promising way to leverage the powerful pre-trained image LLM. In this work, we explore the limitations of the existing compression strategies for building a training-free video LLM. The findings lead to our method TS-LLaVA, which constructs visual tokens through a Thumbnail-and-Sampling strategy. Given a video, we select few equidistant frames from all input frames to construct a Thumbnail image as a detailed visual cue, complemented by Sampled visual tokens from all input frames. Our method establishes the new state-of-the-art performance among training-free video LLMs on various benchmarks. Notably, our 34B model outperforms GPT-4V on the MVBench benchmark, and achieves performance comparable to the 72B training-based video LLM, Video-LLaMA2, on the challenging MLVU benchmark. Code is available at https://github.com/tingyu215/TS-LLaVA.
Latest developments in Large Multimodal Models (LMMs) have broadened their capabilities to include video understanding. Specifically, Text-to-video (T2V) models have made significant progress in quality, comprehension, and duration, excelling at creating videos from simple textual prompts. Yet, they still frequently produce hallucinated content that clearly signals the video is AI-generated. We introduce ViBe: a large-scale Text-to-Video Benchmark of hallucinated videos from T2V models. We identify five major types of hallucination: Vanishing Subject, Numeric Variability, Temporal Dysmorphia, Omission Error, and Physical Incongruity. Using 10 open-source T2V models, we developed the first large-scale dataset of hallucinated videos, comprising 3,782 videos annotated by humans into these five categories. ViBe offers a unique resource for evaluating the reliability of T2V models and provides a foundation for improving hallucination detection and mitigation in video generation. We establish classification as a baseline and present various ensemble classifier configurations, with the TimeSFormer + CNN combination yielding the best performance, achieving 0.345 accuracy and 0.342 F1 score. This benchmark aims to drive the development of robust T2V models that produce videos more accurately aligned with input prompts.
Recent breakthroughs in Multimodal Large Language Models (MLLMs) have gained significant recognition within the deep learning community, where the fusion of the Video Foundation Models (VFMs) and Large Language Models(LLMs) has proven instrumental in constructing robust video understanding systems, effectively surmounting constraints associated with predefined visual tasks. These sophisticated MLLMs exhibit remarkable proficiency in comprehending videos, swiftly attaining unprecedented performance levels across diverse benchmarks. However, their operation demands substantial memory and computational resources, underscoring the continued importance of traditional models in video comprehension tasks. In this paper, we introduce a novel learning paradigm termed MLLM4WTAL. This paradigm harnesses the potential of MLLM to offer temporal action key semantics and complete semantic priors for conventional Weakly-supervised Temporal Action Localization (WTAL) methods. MLLM4WTAL facilitates the enhancement of WTAL by leveraging MLLM guidance. It achieves this by integrating two distinct modules: Key Semantic Matching (KSM) and Complete Semantic Reconstruction (CSR). These modules work in tandem to effectively address prevalent issues like incomplete and over-complete outcomes common in WTAL methods. Rigorous experiments are conducted to validate the efficacy of our proposed approach in augmenting the performance of various heterogeneous WTAL models.
Video question-answering (QA) is a core task in video understanding. Evaluating the quality of video QA and video caption data quality for training video large language models (VideoLLMs) is an essential challenge. Although various methods have been proposed for assessing video caption quality, there remains a lack of dedicated evaluation methods for Video QA. To address this gap, we introduce EVQAScore, a reference-free method that leverages keyword extraction to assess both video caption and video QA data quality. Additionally, we incorporate frame sampling and rescaling techniques to enhance the efficiency and robustness of our evaluation, this enables our score to evaluate the quality of extremely long videos. Our approach achieves state-of-the-art (SOTA) performance (32.8 for Kendall correlation and 42.3 for Spearman correlation, 4.7 and 5.9 higher than the previous method PAC-S++) on the VATEX-EVAL benchmark for video caption evaluation. Furthermore, by using EVQAScore for data selection, we achieved SOTA results with only 12.5\% of the original data volume, outperforming the previous SOTA method PAC-S and 100\% of data.
To address the challenges of high computational costs and long-distance dependencies in exist ing video understanding methods, such as CNNs and Transformers, this work introduces RWKV to the video domain in a novel way. We propose a LSTM CrossRWKV (LCR) framework, designed for spatiotemporal representation learning to tackle the video understanding task. Specifically, the proposed linear complexity LCR incorporates a novel Cross RWKV gate to facilitate interaction be tween current frame edge information and past features, enhancing the focus on the subject through edge features and globally aggregating inter-frame features over time. LCR stores long-term mem ory for video processing through an enhanced LSTM recurrent execution mechanism. By leveraging the Cross RWKV gate and recurrent execution, LCR effectively captures both spatial and temporal features. Additionally, the edge information serves as a forgetting gate for LSTM, guiding long-term memory management.Tube masking strategy reduces redundant information in food and reduces overfitting.These advantages enable LSTM CrossRWKV to set a new benchmark in video under standing, offering a scalable and efficient solution for comprehensive video analysis. All code and models are publicly available.
The rapid development of Multimodal Large Language Models (MLLMs) has expanded their capabilities from image comprehension to video understanding. However, most of these MLLMs focus primarily on offline video comprehension, necessitating extensive processing of all video frames before any queries can be made. This presents a significant gap compared to the human ability to watch, listen, think, and respond to streaming inputs in real time, highlighting the limitations of current MLLMs. In this paper, we introduce StreamingBench, the first comprehensive benchmark designed to evaluate the streaming video understanding capabilities of MLLMs. StreamingBench assesses three core aspects of streaming video understanding: (1) real-time visual understanding, (2) omni-source understanding, and (3) contextual understanding. The benchmark consists of 18 tasks, featuring 900 videos and 4,500 human-curated QA pairs. Each video features five questions presented at different time points to simulate a continuous streaming scenario. We conduct experiments on StreamingBench with 13 open-source and proprietary MLLMs and find that even the most advanced proprietary MLLMs like Gemini 1.5 Pro and GPT-4o perform significantly below human-level streaming video understanding capabilities. We hope our work can facilitate further advancements for MLLMs, empowering them to approach human-level video comprehension and interaction in more realistic scenarios.
Video summarization techniques have been proven to improve the overall user experience when it comes to accessing and comprehending video content. If the user's preference is known, video summarization can identify significant information or relevant content from an input video, aiding them in obtaining the necessary information or determining their interest in watching the original video. Adapting video summarization to various types of video and user preferences requires significant training data and expensive human labeling. To facilitate such research, we proposed a new benchmark for video summarization that captures various user preferences. Also, we present a pipeline called Video Summarization with Language (VSL) for user-preferred video summarization that is based on pre-trained visual language models (VLMs) to avoid the need to train a video summarization system on a large training dataset. The pipeline takes both video and closed captioning as input and performs semantic analysis at the scene level by converting video frames into text. Subsequently, the user's genre preference was used as the basis for selecting the pertinent textual scenes. The experimental results demonstrate that our proposed pipeline outperforms current state-of-the-art unsupervised video summarization models. We show that our method is more adaptable across different datasets compared to supervised query-based video summarization models. In the end, the runtime analysis demonstrates that our pipeline is more suitable for practical use when scaling up the number of user preferences and videos.
The past year has witnessed the significant advancement of video-based large language models. However, the challenge of developing a unified model for both short and long video understanding remains unresolved. Most existing video LLMs cannot handle hour-long videos, while methods custom for long videos tend to be ineffective for shorter videos and images. In this paper, we identify the key issue as the redundant content in videos. To address this, we propose a novel pooling strategy that simultaneously achieves token compression and instruction-aware visual feature aggregation. Our model is termed Prompt-guided Pooling LLaVA, or PPLLaVA for short. Specifically, PPLLaVA consists of three core components: the CLIP-based visual-prompt alignment that extracts visual information relevant to the user's instructions, the prompt-guided pooling that compresses the visual sequence to arbitrary scales using convolution-style pooling, and the clip context extension designed for lengthy prompt common in visual dialogue. Moreover, our codebase also integrates the most advanced video Direct Preference Optimization (DPO) and visual interleave training. Extensive experiments have validated the performance of our model. With superior throughput and only 1024 visual context, PPLLaVA achieves better results on image benchmarks as a video LLM, while achieving state-of-the-art performance across various video benchmarks, excelling in tasks ranging from caption generation to multiple-choice questions, and handling video lengths from seconds to hours. Codes have been available at https://github.com/farewellthree/PPLLaVA.
As the scale of data and models for video understanding rapidly expand, handling long-form video input in transformer-based models presents a practical challenge. Rather than resorting to input sampling or token dropping, which may result in information loss, token merging shows promising results when used in collaboration with transformers. However, the application of token merging for long-form video processing is not trivial. We begin with the premise that token merging should not rely solely on the similarity of video tokens; the saliency of tokens should also be considered. To address this, we explore various video token merging strategies for long-form video classification, starting with a simple extension of image token merging, moving to region-concentrated merging, and finally proposing a learnable video token merging (VTM) algorithm that dynamically merges tokens based on their saliency. Extensive experimental results show that we achieve better or comparable performances on the LVU, COIN, and Breakfast datasets. Moreover, our approach significantly reduces memory costs by 84% and boosts throughput by approximately 6.89 times compared to baseline algorithms.
Graph based representation has been widely used in modelling spatio-temporal relationships in video understanding. Although effective, existing graph-based approaches focus on capturing the human-object relationships while ignoring fine-grained semantic properties of the action components. These semantic properties are crucial for understanding the current situation, such as where does the action takes place, what tools are used and functional properties of the objects. In this work, we propose a graph-based representation called Situational Scene Graph (SSG) to encode both human-object relationships and the corresponding semantic properties. The semantic details are represented as predefined roles and values inspired by situation frame, which is originally designed to represent a single action. Based on our proposed representation, we introduce the task of situational scene graph generation and propose a multi-stage pipeline Interactive and Complementary Network (InComNet) to address the task. Given that the existing datasets are not applicable to the task, we further introduce a SSG dataset whose annotations consist of semantic role-value frames for human, objects and verb predicates of human-object relations. Finally, we demonstrate the effectiveness of our proposed SSG representation by testing on different downstream tasks. Experimental results show that the unified representation can not only benefit predicate classification and semantic role-value classification, but also benefit reasoning tasks on human-centric situation understanding. We will release the code and the dataset soon.
Existing benchmarks often highlight the remarkable performance achieved by state-of-the-art Multimodal Foundation Models (MFMs) in leveraging temporal context for video understanding. However, how well do the models truly perform visual temporal reasoning? Our study of existing benchmarks shows that this capability of MFMs is likely overestimated as many questions can be solved by using a single, few, or out-of-order frames. To systematically examine current visual temporal reasoning tasks, we propose three principles with corresponding metrics: (1) Multi-Frame Gain, (2) Frame Order Sensitivity, and (3) Frame Information Disparity. Following these principles, we introduce TOMATO, Temporal Reasoning Multimodal Evaluation, a novel benchmark crafted to rigorously assess MFMs' temporal reasoning capabilities in video understanding. TOMATO comprises 1,484 carefully curated, human-annotated questions spanning six tasks (i.e., action count, direction, rotation, shape & trend, velocity & frequency, and visual cues), applied to 1,417 videos, including 805 self-recorded and -generated videos, that encompass human-centric, real-world, and simulated scenarios. Our comprehensive evaluation reveals a human-model performance gap of 57.3% with the best-performing model. Moreover, our in-depth analysis uncovers more fundamental limitations beyond this gap in current MFMs. While they can accurately recognize events in isolated frames, they fail to interpret these frames as a continuous sequence. We believe TOMATO will serve as a crucial testbed for evaluating the next-generation MFMs and as a call to the community to develop AI systems capable of comprehending human world dynamics through the video modality.
Egocentric vision captures the scene from the point of view of the camera wearer, while exocentric vision captures the overall scene context. Jointly modeling ego and exo views is crucial to developing next-generation AI agents. The community has regained interest in the field of egocentric vision. While the third-person view and first-person have been thoroughly investigated, very few works aim to study both synchronously. Exocentric videos contain many relevant signals that are transferrable to egocentric videos. This paper provides a timely overview of works combining egocentric and exocentric visions, a very new but promising research topic. We describe in detail the datasets and present a survey of the key applications of ego-exo joint learning, where we identify the most recent advances. With the presentation of the current status of the progress, we believe this short but timely survey will be valuable to the broad video-understanding community, particularly when multi-view modeling is critical.
The growing demand for surveillance in public spaces presents significant challenges due to the shortage of human resources. Current AI-based video surveillance systems heavily rely on core computer vision models that require extensive finetuning, which is particularly difficult in surveillance settings due to limited datasets and difficult setting (viewpoint, low quality, etc.). In this work, we propose leveraging Large Vision-Language Models (LVLMs), known for their strong zero and few-shot generalization, to tackle video understanding tasks in surveillance. Specifically, we explore VideoLLaMA2, a state-of-the-art LVLM, and an improved token-level sampling method, Self-Reflective Sampling (Self-ReS). Our experiments on the UCF-Crime dataset show that VideoLLaMA2 represents a significant leap in zero-shot performance, with 20% boost over the baseline. Self-ReS additionally increases zero-shot action recognition performance to 44.6%. These results highlight the potential of LVLMs, paired with improved sampling techniques, for advancing surveillance video analysis in diverse scenarios.
Understanding long-form video content presents significant challenges due to its temporal complexity and the substantial computational resources required. In this work, we propose an agent-based approach to enhance both the efficiency and effectiveness of long-form video understanding by utilizing large language models (LLMs) and their tool-harnessing ability. A key aspect of our method is query-adaptive frame sampling, which leverages the reasoning capabilities of LLMs to process only the most relevant frames in real-time, and addresses an important limitation of existing methods which typically involve sampling redundant or irrelevant frames. To enhance the reasoning abilities of our video-understanding agent, we leverage the self-reflective capabilities of LLMs to provide verbal reinforcement to the agent, which leads to improved performance while minimizing the number of frames accessed. We evaluate our method across several video understanding benchmarks and demonstrate that not only it enhances state-of-the-art performance but also improves efficiency by reducing the number of frames sampled.
Videos are often used to learn or extract the necessary information to complete tasks in ways different than what text and static imagery alone can provide. However, many existing agent benchmarks neglect long-context video understanding, instead focusing on text or static image inputs. To bridge this gap, we introduce VideoWebArena (VideoWA), a benchmark for evaluating the capabilities of long-context multimodal agents for video understanding. VideoWA consists of 2,021 web agent tasks based on manually crafted video tutorials, which total almost four hours of content. For our benchmark, we define a taxonomy of long-context video-based agent tasks with two main areas of focus: skill retention and factual retention. While skill retention tasks evaluate whether an agent can use a given human demonstration to complete a task efficiently, the factual retention task evaluates whether an agent can retrieve instruction-relevant information from a video to complete a task. We find that the best model achieves 13.3% success on factual retention tasks and 45.8% on factual retention QA pairs, far below human performance at 73.9% and 79.3%, respectively. On skill retention tasks, long-context models perform worse with tutorials than without, exhibiting a 5% performance decrease in WebArena tasks and a 10.3% decrease in VisualWebArena tasks. Our work highlights the need to improve the agentic abilities of long-context multimodal models and provides a testbed for future development with long-context video agents.
Multimodal Large Language Models (MLLMs) have demonstrated impressive performance in short video understanding. However, understanding long-form videos still remains challenging for MLLMs. This paper proposes TimeSuite, a collection of new designs to adapt the existing short-form video MLLMs for long video understanding, including a simple yet efficient framework to process long video sequence, a high-quality video dataset for grounded tuning of MLLMs, and a carefully-designed instruction tuning task to explicitly incorporate the grounding supervision in the traditional QA format. Specifically, based on VideoChat, we propose our long-video MLLM, coined as VideoChat-T, by implementing a token shuffling to compress long video tokens and introducing Temporal Adaptive Position Encoding (TAPE) to enhance the temporal awareness of visual representation. Meanwhile, we introduce the TimePro, a comprehensive grounding-centric instruction tuning dataset composed of 9 tasks and 349k high-quality grounded annotations. Notably, we design a new instruction tuning task type, called Temporal Grounded Caption, to peform detailed video descriptions with the corresponding time stamps prediction. This explicit temporal location prediction will guide MLLM to correctly attend on the visual content when generating description, and thus reduce the hallucination risk caused by the LLMs. Experimental results demonstrate that our TimeSuite provides a successful solution to enhance the long video understanding capability of short-form MLLM, achieving improvement of 5.6% and 6.8% on the benchmarks of Egoschema and VideoMME, respectively. In addition, VideoChat-T exhibits robust zero-shot temporal grounding capabilities, significantly outperforming the existing state-of-the-art MLLMs. After fine-tuning, it performs on par with the traditional supervised expert models.
Recent years have witnessed a significant interest in developing large multimodal models (LMMs) capable of performing various visual reasoning and understanding tasks. This has led to the introduction of multiple LMM benchmarks to evaluate LMMs on different tasks. However, most existing LMM evaluation benchmarks are predominantly English-centric. In this work, we develop a comprehensive LMM evaluation benchmark for the Arabic language to represent a large population of over 400 million speakers. The proposed benchmark, named CAMEL-Bench, comprises eight diverse domains and 38 sub-domains including, multi-image understanding, complex visual perception, handwritten document understanding, video understanding, medical imaging, plant diseases, and remote sensing-based land use understanding to evaluate broad scenario generalizability. Our CAMEL-Bench comprises around 29,036 questions that are filtered from a larger pool of samples, where the quality is manually verified by native speakers to ensure reliable model assessment. We conduct evaluations of both closed-source, including GPT-4 series, and open-source LMMs. Our analysis reveals the need for substantial improvement, especially among the best open-source models, with even the closed-source GPT-4o achieving an overall score of 62%. Our benchmark and evaluation scripts are open-sourced.
Multimodal Large Language Models (MLLMs) have shown promising progress in understanding and analyzing video content. However, processing long videos remains a significant challenge constrained by LLM's context size. To address this limitation, we propose LongVU, a spatiotemporal adaptive compression mechanism thats reduces the number of video tokens while preserving visual details of long videos. Our idea is based on leveraging cross-modal query and inter-frame dependencies to adaptively reduce temporal and spatial redundancy in videos. Specifically, we leverage DINOv2 features to remove redundant frames that exhibit high similarity. Then we utilize text-guided cross-modal query for selective frame feature reduction. Further, we perform spatial token reduction across frames based on their temporal dependencies. Our adaptive compression strategy effectively processes a large number of frames with little visual information loss within given context length. Our LongVU consistently surpass existing methods across a variety of video understanding benchmarks, especially on hour-long video understanding tasks such as VideoMME and MLVU. Given a light-weight LLM, our LongVU also scales effectively into a smaller size with state-of-the-art video understanding performance.
Video understanding has become increasingly important with the rise of multi-modality applications. Understanding continuous video poses considerable challenges due to the fast expansion of streaming video, which contains multi-scale and untrimmed events. We introduce a novel system, C-VUE, to overcome these issues through adaptive state modeling. C-VUE has three key designs. The first is a long-range history modeling technique that uses a video-aware approach to retain historical video information. The second is a spatial redundancy reduction technique, which enhances the efficiency of history modeling based on temporal relations. The third is a parallel training structure that incorporates the frame-weighted loss to understand multi-scale events in long videos. Our C-VUE offers high accuracy and efficiency. It runs at speeds >30 FPS on typical edge devices and outperforms all baselines in accuracy. Moreover, applying C-VUE to a video foundation model as a video encoder in our case study resulted in a 0.46-point enhancement (on a 5-point scale) on the in-distribution dataset, and an improvement ranging from 1.19\% to 4\% on zero-shot datasets.