Egocentric videos provide a unique perspective into individuals' daily experiences, yet their unstructured nature presents challenges for perception. In this paper, we introduce AMEGO, a novel approach aimed at enhancing the comprehension of very-long egocentric videos. Inspired by the human's ability to maintain information from a single watching, AMEGO focuses on constructing a self-contained representations from one egocentric video, capturing key locations and object interactions. This representation is semantic-free and facilitates multiple queries without the need to reprocess the entire visual content. Additionally, to evaluate our understanding of very-long egocentric videos, we introduce the new Active Memories Benchmark (AMB), composed of more than 20K of highly challenging visual queries from EPIC-KITCHENS. These queries cover different levels of video reasoning (sequencing, concurrency and temporal grounding) to assess detailed video understanding capabilities. We showcase improved performance of AMEGO on AMB, surpassing other video QA baselines by a substantial margin.
Recently, integrating visual foundation models into large language models (LLMs) to form video understanding systems has attracted widespread attention. Most of the existing models compress diverse semantic information within the whole video and feed it into LLMs for content comprehension. While this method excels in short video understanding, it may result in a blend of multiple event information in long videos due to coarse compression, which causes information redundancy. Consequently, the semantics of key events might be obscured within the vast information that hinders the model's understanding capabilities. To address this issue, we propose a Hierarchical Event-based Memory-enhanced LLM (HEM-LLM) for better understanding of long videos. Firstly, we design a novel adaptive sequence segmentation scheme to divide multiple events within long videos. In this way, we can perform individual memory modeling for each event to establish intra-event contextual connections, thereby reducing information redundancy. Secondly, while modeling current event, we compress and inject the information of the previous event to enhance the long-term inter-event dependencies in videos. Finally, we perform extensive experiments on various video understanding tasks and the results show that our model achieves state-of-the-art performances.
We introduce VidLPRO, a novel video-language (VL) pre-training framework designed specifically for robotic and laparoscopic surgery. While existing surgical VL models primarily rely on contrastive learning, we propose a more comprehensive approach to capture the intricate temporal dynamics and align video with language. VidLPRO integrates video-text contrastive learning, video-text matching, and masked language modeling objectives to learn rich VL representations. To support this framework, we present GenSurg+, a carefully curated dataset derived from GenSurgery, comprising 17k surgical video clips paired with captions generated by GPT-4 using transcripts extracted by the Whisper model. This dataset addresses the need for large-scale, high-quality VL data in the surgical domain. Extensive experiments on benchmark datasets, including Cholec80 and AutoLaparo, demonstrate the efficacy of our approach. VidLPRO achieves state-of-the-art performance in zero-shot surgical phase recognition, significantly outperforming existing surgical VL models such as SurgVLP and HecVL. Our model demonstrates improvements of up to 21.5\% in accuracy and 15.7% in F1 score, setting a new benchmark in the field. Notably, VidLPRO exhibits robust performance even with single-frame inference, while effectively scaling with increased temporal context. Ablation studies reveal the impact of frame sampling strategies on model performance and computational efficiency. These results underscore VidLPRO's potential as a foundation model for surgical video understanding.
Multimodal Large Language Models (MLLMs) have significantly improved performance across various image-language applications. Recently, there has been a growing interest in adapting image pre-trained MLLMs for video-related tasks. However, most efforts concentrate on enhancing the vision encoder and projector components, while the core part, Large Language Models (LLMs), remains comparatively under-explored. In this paper, we propose two strategies to enhance the model's capability in video understanding tasks by improving inter-layer attention computation in LLMs. Specifically, the first approach focuses on the enhancement of Rotary Position Embedding (RoPE) with Temporal-Aware Dual RoPE, which introduces temporal position information to strengthen the MLLM's temporal modeling capabilities while preserving the relative position relationships of both visual and text tokens. The second approach involves enhancing the Attention Mask with the Frame-wise Block Causal Attention Mask, a simple yet effective method that broadens visual token interactions within and across video frames while maintaining the causal inference mechanism. Based on these proposed methods, we adapt LLaVA for video understanding tasks, naming it Temporal-Considered LLaVA (TC-LLaVA). Our TC-LLaVA achieves new state-of-the-art performance across various video understanding benchmarks with only supervised fine-tuning (SFT) on video-related datasets.
Recent advancements in large-scale video-language models have shown significant potential for real-time planning and detailed interactions. However, their high computational demands and the scarcity of annotated datasets limit their practicality for academic researchers. In this work, we introduce VideoLLaMB, a novel framework that utilizes temporal memory tokens within bridge layers to allow for the encoding of entire video sequences alongside historical visual data, effectively preserving semantic continuity and enhancing model performance across various tasks. This approach includes recurrent memory tokens and a SceneTilling algorithm, which segments videos into independent semantic units to preserve semantic integrity. Empirically, VideoLLaMB significantly outstrips existing video-language models, demonstrating a 5.5 points improvement over its competitors across three VideoQA benchmarks, and 2.06 points on egocentric planning. Comprehensive results on the MVBench show that VideoLLaMB-7B achieves markedly better results than previous 7B models of same LLM. Remarkably, it maintains robust performance as PLLaVA even as video length increases up to 8 times. Besides, the frame retrieval results on our specialized Needle in a Video Haystack (NIAVH) benchmark, further validate VideoLLaMB's prowess in accurately identifying specific frames within lengthy videos. Our SceneTilling algorithm also enables the generation of streaming video captions directly, without necessitating additional training. In terms of efficiency, VideoLLaMB, trained on 16 frames, supports up to 320 frames on a single Nvidia A100 GPU with linear GPU memory scaling, ensuring both high performance and cost-effectiveness, thereby setting a new foundation for long-form video-language models in both academic and practical applications.
In recent years, unmanned aerial vehicles (UAVs) have played an increasingly crucial role in supporting disaster emergency response efforts by analyzing aerial images. While current deep-learning models focus on improving accuracy, they often overlook the limited computing resources of UAVs. This study recognizes the imperative for real-time data processing in disaster response scenarios and introduces a lightweight and efficient approach for aerial video understanding. Our methodology identifies redundant portions within the video through policy networks and eliminates this excess information using frame compression techniques. Additionally, we introduced the concept of a `station point,' which leverages future information in the sequential policy network, thereby enhancing accuracy. To validate our method, we employed the wildfire FLAME dataset. Compared to the baseline, our approach reduces computation costs by more than 13 times while boosting accuracy by 3$\%$. Moreover, our method can intelligently select salient frames from the video, refining the dataset. This feature enables sophisticated models to be effectively trained on a smaller dataset, significantly reducing the time spent during the training process.
Expanding the long-context capabilities of Multi-modal Large Language Models~(MLLMs) is crucial for video understanding, high-resolution image understanding, and multi-modal agents. This involves a series of systematic optimizations, including model architecture, data construction and training strategy, particularly addressing challenges such as \textit{degraded performance with more images} and \textit{high computational costs}. In this paper, we adapt the model architecture to a hybrid of Mamba and Transformer blocks, approach data construction with both temporal and spatial dependencies among multiple images and employ a progressive training strategy. The released model \textbf{LongLLaVA}~(\textbf{Long}-Context \textbf{L}arge \textbf{L}anguage \textbf{a}nd \textbf{V}ision \textbf{A}ssistant) is the first hybrid MLLM, which achieved a better balance between efficiency and effectiveness. LongLLaVA not only achieves competitive results across various benchmarks, but also maintains high throughput and low memory consumption. Especially, it could process nearly a thousand images on a single A100 80GB GPU, showing promising application prospects for a wide range of tasks.
In recent years, the parameters of backbones of Video Understanding tasks continue to increase and even reach billion-level. Whether fine-tuning a specific task on the Video Foundation Model or pre-training the model designed for the specific task, incurs a lot of overhead. How to make these models play other values than their own tasks becomes a worthy question. Multi-Task Learning(MTL) makes the visual task acquire the rich shareable knowledge from other tasks while joint training. It is fully explored in Image Recognition tasks especially dense predict tasks. Nevertheless, it is rarely used in video domain due to the lack of multi-labels video data. In this paper, a heterogenous data video multi-task prompt learning (VMTL) method is proposed to address above problem. It's different from it in image domain, a Double-Layers Mapper(DLM) is proposed to extract the shareable knowledge into visual promptS and align it with representation of primary task. Extensive experiments prove that our DLM-VMTL performs better than baselines on 6 different video understanding tasks and 11 datasets.
Beginning with VisualGLM and CogVLM, we are continuously exploring VLMs in pursuit of enhanced vision-language fusion, efficient higher-resolution architecture, and broader modalities and applications. Here we propose the CogVLM2 family, a new generation of visual language models for image and video understanding including CogVLM2, CogVLM2-Video and GLM-4V. As an image understanding model, CogVLM2 inherits the visual expert architecture with improved training recipes in both pre-training and post-training stages, supporting input resolution up to $1344 \times 1344$ pixels. As a video understanding model, CogVLM2-Video integrates multi-frame input with timestamps and proposes automated temporal grounding data construction. Notably, CogVLM2 family has achieved state-of-the-art results on benchmarks like MMBench, MM-Vet, TextVQA, MVBench and VCGBench. All models are open-sourced in https://github.com/THUDM/CogVLM2 and https://github.com/THUDM/GLM-4, contributing to the advancement of the field.
Rapid advancements have been made in extending Large Language Models (LLMs) to Large Multi-modal Models (LMMs). However, extending input modality of LLMs to video data remains a challenging endeavor, especially for long videos. Due to insufficient access to large-scale high-quality video data and the excessive compression of visual features, current methods exhibit limitations in effectively processing long videos. In this paper, we introduce Kangaroo, a powerful Video LMM aimed at addressing these challenges. Confronted with issue of inadequate training data, we develop a data curation system to build a large-scale dataset with high-quality annotations for vision-language pre-training and instruction tuning. In addition, we design a curriculum training pipeline with gradually increasing resolution and number of input frames to accommodate long videos. Evaluation results demonstrate that, with 8B parameters, Kangaroo achieves state-of-the-art performance across a variety of video understanding benchmarks while exhibiting competitive results on others. Particularly, on benchmarks specialized for long videos, Kangaroo excels some larger models with over 10B parameters and proprietary models.
Multi-modal large language models (MLLMs) have demonstrated considerable potential across various downstream tasks that require cross-domain knowledge. MLLMs capable of processing videos, known as Video-MLLMs, have attracted broad interest in video-language understanding. However, videos, especially long videos, contain more visual tokens than images, making them difficult for LLMs to process. Existing works either downsample visual features or extend the LLM context size, risking the loss of high-resolution information or slowing down inference speed. To address these limitations, we apply cross-attention layers in the intermediate projector between the visual encoder and the large language model (LLM). As the naive cross-attention mechanism is insensitive to temporal order, we further introduce causal cross-attention masks (CCAMs) within the cross-attention layers. This Video-MLLM, named Video-CCAM, is trained in a straightforward two-stage fashion: feature alignment and visual instruction tuning. We develop several Video-CCAM models based on LLMs of different sizes (4B, 9B, and 14B). Video-CCAM proves to be a robust Video-MLLM and shows outstanding performance from short videos to long ones. Among standard video benchmarks like MVBench and VideoChatGPT-QA, Video-CCAM shows outstanding performances (1st/2nd/3rd in MVBench and TGIF-QA, 2nd/3rd/4th in MSVD-QA, MSRVTT-QA, and ActivityNet-QA). In benchmarks encompassing long videos, Video-CCAM models can be directly adapted to long video understanding and still achieve exceptional scores despite being trained solely with images and 16-frame videos. Using 96 frames (6$\times$ the training number of frames), Video-CCAM models rank 1st/2nd/3rd in VideoVista and 1st/2nd/4th in MLVU among all open-source Video-MLLMs, respectively. The code is publicly available in \url{https://github.com/QQ-MM/Video-CCAM}.
The explosive growth of videos on streaming media platforms has underscored the urgent need for effective video quality assessment (VQA) algorithms to monitor and perceptually optimize the quality of streaming videos. However, VQA remains an extremely challenging task due to the diverse video content and the complex spatial and temporal distortions, thus necessitating more advanced methods to address these issues. Nowadays, large multimodal models (LMMs), such as GPT-4V, have exhibited strong capabilities for various visual understanding tasks, motivating us to leverage the powerful multimodal representation ability of LMMs to solve the VQA task. Therefore, we propose the first Large Multi-Modal Video Quality Assessment (LMM-VQA) model, which introduces a novel spatiotemporal visual modeling strategy for quality-aware feature extraction. Specifically, we first reformulate the quality regression problem into a question and answering (Q&A) task and construct Q&A prompts for VQA instruction tuning. Then, we design a spatiotemporal vision encoder to extract spatial and temporal features to represent the quality characteristics of videos, which are subsequently mapped into the language space by the spatiotemporal projector for modality alignment. Finally, the aligned visual tokens and the quality-inquired text tokens are aggregated as inputs for the large language model (LLM) to generate the quality score and level. Extensive experiments demonstrate that LMM-VQA achieves state-of-the-art performance across five VQA benchmarks, exhibiting an average improvement of $5\%$ in generalization ability over existing methods. Furthermore, due to the advanced design of the spatiotemporal encoder and projector, LMM-VQA also performs exceptionally well on general video understanding tasks, further validating its effectiveness. Our code will be released at https://github.com/Sueqk/LMM-VQA.
Exploiting both audio and visual modalities for video classification is a challenging task, as the existing methods require large model architectures, leading to high computational complexity and resource requirements. Smaller architectures, on the other hand, struggle to achieve optimal performance. In this paper, we propose Attend-Fusion, an audio-visual (AV) fusion approach that introduces a compact model architecture specifically designed to capture intricate audio-visual relationships in video data. Through extensive experiments on the challenging YouTube-8M dataset, we demonstrate that Attend-Fusion achieves an F1 score of 75.64\% with only 72M parameters, which is comparable to the performance of larger baseline models such as Fully-Connected Late Fusion (75.96\% F1 score, 341M parameters). Attend-Fusion achieves similar performance to the larger baseline model while reducing the model size by nearly 80\%, highlighting its efficiency in terms of model complexity. Our work demonstrates that the Attend-Fusion model effectively combines audio and visual information for video classification, achieving competitive performance with significantly reduced model size. This approach opens new possibilities for deploying high-performance video understanding systems in resource-constrained environments across various applications.
The recent wave of foundation models has witnessed tremendous success in computer vision (CV) and beyond, with the segment anything model (SAM) having sparked a passion for exploring task-agnostic visual foundation models. Empowered by its remarkable zero-shot generalization, SAM is currently challenging numerous traditional paradigms in CV, delivering extraordinary performance not only in various image segmentation and multi-modal segmentation (\eg, text-to-mask) tasks, but also in the video domain. Additionally, the latest released SAM 2 is once again sparking research enthusiasm in the realm of promptable visual segmentation for both images and videos. However, existing surveys mainly focus on SAM in various image processing tasks, a comprehensive and in-depth review in the video domain is notably absent. To address this gap, this work conducts a systematic review on SAM for videos in the era of foundation models. As the first to review the progress of SAM for videos, this work focuses on its applications to various tasks by discussing its recent advances, and innovation opportunities of developing foundation models on broad applications. We begin with a brief introduction to the background of SAM and video-related research domains. Subsequently, we present a systematic taxonomy that categorizes existing methods into three key areas: video understanding, video generation, and video editing, analyzing and summarizing their advantages and limitations. Furthermore, comparative results of SAM-based and current state-of-the-art methods on representative benchmarks, as well as insightful analysis are offered. Finally, we discuss the challenges faced by current research and envision several future research directions in the field of SAM for video and beyond.
Long-context capability is critical for multi-modal foundation models, especially for long video understanding. We introduce LongVILA, a full-stack solution for long-context visual-language models by co-designing the algorithm and system. For model training, we upgrade existing VLMs to support long video understanding by incorporating two additional stages, i.e., long context extension and long video supervised fine-tuning. However, training on long video is computationally and memory intensive. We introduce the long-context Multi-Modal Sequence Parallelism (MM-SP) system that efficiently parallelizes long video training and inference, enabling 2M context length training on 256 GPUs without any gradient checkpointing. LongVILA efficiently extends the number of video frames of VILA from 8 to 2048, achieving 99.8% accuracy in 6,000-frame (more than 1 million tokens) video needle-in-a-haystack. LongVILA-7B demonstrates strong accuracy on 9 popular video benchmarks, e.g. 65.1% VideoMME with subtitle. Besides, MM-SP is 2.1x - 5.7x faster than ring style sequence parallelism and 1.1x - 1.4x faster than Megatron with a hybrid context and tensor parallelism. Moreover, it seamlessly integrates with Hugging Face Transformers.
In recent years, video action recognition, as a fundamental task in the field of video understanding, has been deeply explored by numerous researchers.Most traditional video action recognition methods typically involve converting videos into three-dimensional data that encapsulates both spatial and temporal information, subsequently leveraging prevalent image understanding models to model and analyze these data. However,these methods have significant drawbacks. Firstly, when delving into video action recognition tasks, image understanding models often need to be adapted accordingly in terms of model architecture and preprocessing for these spatiotemporal tasks; Secondly, dealing with high-dimensional data often poses greater challenges and incurs higher time costs compared to its lower-dimensional counterparts.To bridge the gap between image-understanding and video-understanding tasks while simplifying the complexity of video comprehension, we introduce a novel video representation architecture, Flatten, which serves as a plug-and-play module that can be seamlessly integrated into any image-understanding network for efficient and effective 3D temporal data modeling.Specifically, by applying specific flattening operations (e.g., row-major transform), 3D spatiotemporal data is transformed into 2D spatial information, and then ordinary image understanding models are used to capture temporal dynamic and spatial semantic information, which in turn accomplishes effective and efficient video action recognition. Extensive experiments on commonly used datasets (Kinetics-400, Something-Something v2, and HMDB-51) and three classical image classification models (Uniformer, SwinV2, and ResNet), have demonstrated that embedding Flatten provides a significant performance improvements over original model.
Video Moment Retrieval, which aims to locate in-context video moments according to a natural language query, is an essential task for cross-modal grounding. Existing methods focus on enhancing the cross-modal interactions between all moments and the textual description for video understanding. However, constantly interacting with all locations is unreasonable because of uneven semantic distribution across the timeline and noisy visual backgrounds. This paper proposes a cross-modal Context Denoising Network (CDNet) for accurate moment retrieval by disentangling complex correlations and denoising irrelevant dynamics.Specifically, we propose a query-guided semantic disentanglement (QSD) to decouple video moments by estimating alignment levels according to the global and fine-grained correlation. A Context-aware Dynamic Denoisement (CDD) is proposed to enhance understanding of aligned spatial-temporal details by learning a group of query-relevant offsets. Extensive experiments on public benchmarks demonstrate that the proposed CDNet achieves state-of-the-art performances.
Online video understanding often relies on individual frames, leading to frame-by-frame predictions. Recent advancements such as Online Temporal Action Localization (OnTAL), extend this approach to instance-level predictions. However, existing methods mainly focus on short-term context, neglecting historical information. To address this, we introduce the History-Augmented Anchor Transformer (HAT) Framework for OnTAL. By integrating historical context, our framework enhances the synergy between long-term and short-term information, improving the quality of anchor features crucial for classification and localization. We evaluate our model on both procedural egocentric (PREGO) datasets (EGTEA and EPIC) and standard non-PREGO OnTAL datasets (THUMOS and MUSES). Results show that our model outperforms state-of-the-art approaches significantly on PREGO datasets and achieves comparable or slightly superior performance on non-PREGO datasets, underscoring the importance of leveraging long-term history, especially in procedural and egocentric action scenarios. Code is available at: https://github.com/sakibreza/ECCV24-HAT/
Egocentric videos provide comprehensive contexts for user and scene understanding, spanning multisensory perception to behavioral interaction. We propose Spherical World-Locking (SWL) as a general framework for egocentric scene representation, which implicitly transforms multisensory streams with respect to measurements of head orientation. Compared to conventional head-locked egocentric representations with a 2D planar field-of-view, SWL effectively offsets challenges posed by self-motion, allowing for improved spatial synchronization between input modalities. Using a set of multisensory embeddings on a worldlocked sphere, we design a unified encoder-decoder transformer architecture that preserves the spherical structure of the scene representation, without requiring expensive projections between image and world coordinate systems. We evaluate the effectiveness of the proposed framework on multiple benchmark tasks for egocentric video understanding, including audio-visual active speaker localization, auditory spherical source localization, and behavior anticipation in everyday activities.
Video Large Language Models (Video-LLMs) are flourishing and has advanced many video-language tasks. As a golden testbed, Video Question Answering (VideoQA) plays pivotal role in Video-LLM developing. This work conducts a timely and comprehensive study of Video-LLMs' behavior in VideoQA, aiming to elucidate their success and failure modes, and provide insights towards more human-like video understanding and question answering. Our analyses demonstrate that Video-LLMs excel in VideoQA; they can correlate contextual cues and generate plausible responses to questions about varied video contents. However, models falter in handling video temporality, both in reasoning about temporal content ordering and grounding QA-relevant temporal moments. Moreover, the models behave unintuitively - they are unresponsive to adversarial video perturbations while being sensitive to simple variations of candidate answers and questions. Also, they do not necessarily generalize better. The findings demonstrate Video-LLMs' QA capability in standard condition yet highlight their severe deficiency in robustness and interpretability, suggesting the urgent need on rationales in Video-LLM developing.