Text-to-Video Retrieval (TVR) aims to match videos with corresponding textual queries, yet the continual influx of new video content poses a significant challenge for maintaining system performance over time. In this work, we introduce the first benchmark for Continual Text-to-Video Retrieval (CTVR) to overcome these limitations. Our analysis reveals that current TVR methods based on pre-trained models struggle to retain plasticity when adapting to new tasks, while existing continual learning approaches experience catastrophic forgetting, resulting in semantic misalignment between historical queries and stored video features. To address these challenges, we propose StableFusion, a novel CTVR framework comprising two main components: the Frame Fusion Adapter (FFA), which captures temporal dynamics in video content while preserving model flexibility, and the Task-Aware Mixture-of-Experts (TAME), which maintains consistent semantic alignment between queries across tasks and the stored video features. Comprehensive evaluations on two benchmark datasets under various task settings demonstrate that StableFusion outperforms existing continual learning and TVR methods, achieving superior retrieval performance with minimal degradation on earlier tasks in the context of continuous video streams. Our code is available at: https://github.com/JasonCodeMaker/CTVR
Integrating audio and visual data for training multimodal foundational models remains challenging. We present Audio-Video Vector Alignment (AVVA), which aligns audiovisual (AV) scene content beyond mere temporal synchronization via a Large Language Model (LLM)-based data curation pipeline. Specifically, AVVA scores and selects high-quality training clips using Whisper (speech-based audio foundation model) for audio and DINOv2 for video within a dual-encoder contrastive learning framework. Evaluations on AudioCaps, VALOR, and VGGSound demonstrate that this approach can achieve significant accuracy gains with substantially less curated data. For instance, AVVA yields a 7.6% improvement in top-1 accuracy for audio-to-video retrieval on VGGSound compared to ImageBind, despite training on only 192 hours of carefully filtered data (vs. 5800+ hours). Moreover, an ablation study highlights that trading data quantity for data quality improves performance, yielding respective top-3 accuracy increases of 47.8, 48.4, and 58.0 percentage points on AudioCaps, VALOR, and VGGSound over uncurated baselines. While these results underscore AVVA's data efficiency, we also discuss the overhead of LLM-driven curation and how it may be scaled or approximated in larger domains. Overall, AVVA provides a viable path toward more robust, text-free audiovisual learning with improved retrieval accuracy.
Universal multimodal embedding models play a critical role in tasks such as interleaved image-text retrieval, multimodal RAG, and multimodal clustering. However, our empirical results indicate that existing LMM-based embedding models trained with the standard InfoNCE loss exhibit a high degree of overlap in similarity distribution between positive and negative pairs, making it challenging to distinguish hard negative pairs effectively. To deal with this issue, we propose a simple yet effective framework that dynamically improves the embedding model's representation learning for negative pairs based on their discriminative difficulty. Within this framework, we train a series of models, named LLaVE, and evaluate them on the MMEB benchmark, which covers 4 meta-tasks and 36 datasets. Experimental results show that LLaVE establishes stronger baselines that achieve state-of-the-art (SOTA) performance while demonstrating strong scalability and efficiency. Specifically, LLaVE-2B surpasses the previous SOTA 7B models, while LLaVE-7B achieves a further performance improvement of 6.2 points. Although LLaVE is trained on image-text data, it can generalize to text-video retrieval tasks in a zero-shot manner and achieve strong performance, demonstrating its remarkable potential for transfer to other embedding tasks.
In recent text-video retrieval, the use of additional captions from vision-language models has shown promising effects on the performance. However, existing models using additional captions often have struggled to capture the rich semantics, including temporal changes, inherent in the video. In addition, incorrect information caused by generative models can lead to inaccurate retrieval. To address these issues, we propose a new framework, Narrating the Video (NarVid), which strategically leverages the comprehensive information available from frame-level captions, the narration. The proposed NarVid exploits narration in multiple ways: 1) feature enhancement through cross-modal interactions between narration and video, 2) query-aware adaptive filtering to suppress irrelevant or incorrect information, 3) dual-modal matching score by adding query-video similarity and query-narration similarity, and 4) hard-negative loss to learn discriminative features from multiple perspectives using the two similarities from different views. Experimental results demonstrate that NarVid achieves state-of-the-art performance on various benchmark datasets.
Anticipating future events is crucial for various application domains such as healthcare, smart home technology, and surveillance. Narrative event descriptions provide context-rich information, enhancing a system's future planning and decision-making capabilities. We propose a novel task: $\textit{long-term future narration generation}$, which extends beyond traditional action anticipation by generating detailed narrations of future daily activities. We introduce a visual-language model, ViNa, specifically designed to address this challenging task. ViNa integrates long-term videos and corresponding narrations to generate a sequence of future narrations that predict subsequent events and actions over extended time horizons. ViNa extends existing multimodal models that perform only short-term predictions or describe observed videos by generating long-term future narrations for a broader range of daily activities. We also present a novel downstream application that leverages the generated narrations called future video retrieval to help users improve planning for a task by visualizing the future. We evaluate future narration generation on the largest egocentric dataset Ego4D.
With the rapid growth of multi-modal data from social media, short video platforms, and e-commerce, content-based retrieval has become essential for efficiently searching and utilizing heterogeneous information. Over time, retrieval techniques have evolved from Unimodal Retrieval (UR) to Cross-modal Retrieval (CR) and, more recently, to Composed Multi-modal Retrieval (CMR). CMR enables users to retrieve images or videos by integrating a reference visual input with textual modifications, enhancing search flexibility and precision. This paper provides a comprehensive review of CMR, covering its fundamental challenges, technical advancements, and categorization into supervised, zero-shot, and semi-supervised learning paradigms. We discuss key research directions, including data augmentation, model architecture, and loss optimization in supervised CMR, as well as transformation frameworks and external knowledge integration in zero-shot CMR. Additionally, we highlight the application potential of CMR in composed image retrieval, video retrieval, and person retrieval, which have significant implications for e-commerce, online search, and public security. Given its ability to refine and personalize search experiences, CMR is poised to become a pivotal technology in next-generation retrieval systems. A curated list of related works and resources is available at: https://github.com/kkzhang95/Awesome-Composed-Multi-modal-Retrieval
Transformers have been favored in both uni-modal and multi-modal foundation models for their flexible scalability in attention modules. Consequently, a number of pre-trained Transformer models, e.g., LLaVA, CLIP, and DEIT, are publicly available. Recent research has introduced subquadratic architectures like Mamba, which enables global awareness with linear complexity. Nevertheless, training specialized subquadratic architectures from scratch for certain tasks is both resource-intensive and time-consuming. As a motivator, we explore cross-architecture training to transfer the ready knowledge in existing Transformer models to alternative architecture Mamba, termed TransMamba. Our approach employs a two-stage strategy to expedite training new Mamba models, ensuring effectiveness in across uni-modal and cross-modal tasks. Concerning architecture disparities, we project the intermediate features into an aligned latent space before transferring knowledge. On top of that, a Weight Subcloning and Adaptive Bidirectional distillation method (WSAB) is introduced for knowledge transfer without limitations on varying layer counts. For cross-modal learning, we propose a cross-Mamba module that integrates language awareness into Mamba's visual features, enhancing the cross-modal interaction capabilities of Mamba architecture. Despite using less than 75% of the training data typically required for training from scratch, TransMamba boasts substantially stronger performance across various network architectures and downstream tasks, including image classification, visual question answering, and text-video retrieval. The code will be publicly available.
With the rapid development of AI-generated content (AIGC), the creation of high-quality AI-generated videos has become faster and easier, resulting in the Internet being flooded with all kinds of video content. However, the impact of these videos on the content ecosystem remains largely unexplored. Video information retrieval remains a fundamental approach for accessing video content. Building on the observation that retrieval models often favor AI-generated content in ad-hoc and image retrieval tasks, we investigate whether similar biases emerge in the context of challenging video retrieval, where temporal and visual factors may further influence model behavior. To explore this, we first construct a comprehensive benchmark dataset containing both real and AI-generated videos, along with a set of fair and rigorous metrics to assess bias. This benchmark consists of 13,000 videos generated by two state-of-the-art open-source video generation models. We meticulously design a suite of rigorous metrics to accurately measure this preference, accounting for potential biases arising from the limited frame rate and suboptimal quality of AIGC videos. We then applied three off-the-shelf video retrieval models to perform retrieval tasks on this hybrid dataset. Our findings reveal a clear preference for AI-generated videos in retrieval. Further investigation shows that incorporating AI-generated videos into the training set of retrieval models exacerbates this bias. Unlike the preference observed in image modalities, we find that video retrieval bias arises from both unseen visual and temporal information, making the root causes of video bias a complex interplay of these two factors. To mitigate this bias, we fine-tune the retrieval models using a contrastive learning approach. The results of this study highlight the potential implications of AI-generated videos on retrieval systems.
While Rotary Position Embedding (RoPE) and its variants are widely adopted for their long-context capabilities, the extension of the 1D RoPE to video, with its complex spatio-temporal structure, remains an open challenge. This work first introduces a comprehensive analysis that identifies four key characteristics essential for the effective adaptation of RoPE to video, which have not been fully considered in prior work. As part of our analysis, we introduce a challenging V-NIAH-D (Visual Needle-In-A-Haystack with Distractors) task, which adds periodic distractors into V-NIAH. The V-NIAH-D task demonstrates that previous RoPE variants, lacking appropriate temporal dimension allocation, are easily misled by distractors. Based on our analysis, we introduce \textbf{VideoRoPE}, with a \textit{3D structure} designed to preserve spatio-temporal relationships. VideoRoPE features \textit{low-frequency temporal allocation} to mitigate periodic oscillations, a \textit{diagonal layout} to maintain spatial symmetry, and \textit{adjustable temporal spacing} to decouple temporal and spatial indexing. VideoRoPE consistently surpasses previous RoPE variants, across diverse downstream tasks such as long video retrieval, video understanding, and video hallucination. Our code will be available at \href{https://github.com/Wiselnn570/VideoRoPE}{https://github.com/Wiselnn570/VideoRoPE}.
Video understanding, including video captioning and retrieval, is still a great challenge for video-language models (VLMs). The existing video retrieval and caption benchmarks only include short descriptions, limits their ability of detailed video understanding evaluation. To address this problem, we present CaReBench, a testing benchmark for fine-grained video captioning and retrieval with 1,000 high-quality pairs of videos and human-annotated detailed captions. Uniquely, it provides manually separated spatial annotations and temporal annotations for each video. Based on this design, we introduce two evaluation metrics, ReBias and CapST, specifically tailored for video retrieval and video captioning tasks, respectively. These metrics enable a comprehensive investigation into the spatial and temporal biases inherent in VLMs. In addition, to handle both video retrieval and video captioning tasks in a unified framework, we develop a simple baseline based on a Multimodal Language Model (MLLM). By implementing a two-stage Supervised Fine-Tuning (SFT), we fully unlock the potential of MLLM, enabling it not only to generate detailed video descriptions but also to extract video features. Surprisingly, experimental results demonstrate that, compared to the CLIP-based models designed for retrieval and the popular MLLMs skilled in video captioning, our baseline shows competitive performance in both fine-grained video retrieval and video detailed captioning.
Multimodal representation learning, with contrastive learning, plays an important role in the artificial intelligence domain. As an important subfield, video-language representation learning focuses on learning representations using global semantic interactions between pre-defined video-text pairs. However, to enhance and refine such coarse-grained global interactions, more detailed interactions are necessary for fine-grained multimodal learning. In this study, we introduce a new approach that models video-text as game players using multivariate cooperative game theory to handle uncertainty during fine-grained semantic interactions with diverse granularity, flexible combination, and vague intensity. Specifically, we design the Hierarchical Banzhaf Interaction to simulate the fine-grained correspondence between video clips and textual words from hierarchical perspectives. Furthermore, to mitigate the bias in calculations within Banzhaf Interaction, we propose reconstructing the representation through a fusion of single-modal and cross-modal components. This reconstructed representation ensures fine granularity comparable to that of the single-modal representation, while also preserving the adaptive encoding characteristics of cross-modal representation. Additionally, we extend our original structure into a flexible encoder-decoder framework, enabling the model to adapt to various downstream tasks. Extensive experiments on commonly used text-video retrieval, video-question answering, and video captioning benchmarks, with superior performance, validate the effectiveness and generalization of our method.
Video Corpus Visual Answer Localization (VCVAL) includes question-related video retrieval and visual answer localization in the videos. Specifically, we use text-to-text retrieval to find relevant videos for a medical question based on the similarity of video transcript and answers generated by GPT4. For the visual answer localization, the start and end timestamps of the answer are predicted by the alignments on both visual content and subtitles with queries. For the Query-Focused Instructional Step Captioning (QFISC) task, the step captions are generated by GPT4. Specifically, we provide the video captions generated by the LLaVA-Next-Video model and the video subtitles with timestamps as context, and ask GPT4 to generate step captions for the given medical query. We only submit one run for evaluation and it obtains a F-score of 11.92 and mean IoU of 9.6527.
Video has emerged as a favored multimedia format on the internet. To better gain video contents, a new topic HIREST is presented, including video retrieval, moment retrieval, moment segmentation, and step-captioning. The pioneering work chooses the pre-trained CLIP-based model for video retrieval, and leverages it as a feature extractor for other three challenging tasks solved in a multi-task learning paradigm. Nevertheless, this work struggles to learn the comprehensive cognition of user-preferred content, due to disregarding the hierarchies and association relations across modalities. In this paper, guided by the shallow-to-deep principle, we propose a query-centric audio-visual cognition (QUAG) network to construct a reliable multi-modal representation for moment retrieval, segmentation and step-captioning. Specifically, we first design the modality-synergistic perception to obtain rich audio-visual content, by modeling global contrastive alignment and local fine-grained interaction between visual and audio modalities. Then, we devise the query-centric cognition that uses the deep-level query to perform the temporal-channel filtration on the shallow-level audio-visual representation. This can cognize user-preferred content and thus attain a query-centric audio-visual representation for three tasks. Extensive experiments show QUAG achieves the SOTA results on HIREST. Further, we test QUAG on the query-based video summarization task and verify its good generalization.
Human perception integrates multiple modalities, such as vision, hearing, and language, into a unified understanding of the surrounding reality. While recent multimodal models have achieved significant progress by aligning pairs of modalities via contrastive learning, their solutions are unsuitable when scaling to multiple modalities. These models typically align each modality to a designated anchor without ensuring the alignment of all modalities with each other, leading to suboptimal performance in tasks requiring a joint understanding of multiple modalities. In this paper, we structurally rethink the pairwise conventional approach to multimodal learning and we present the novel Gramian Representation Alignment Measure (GRAM), which overcomes the above-mentioned limitations. GRAM learns and then aligns $n$ modalities directly in the higher-dimensional space in which modality embeddings lie by minimizing the Gramian volume of the $k$-dimensional parallelotope spanned by the modality vectors, ensuring the geometric alignment of all modalities simultaneously. GRAM can replace cosine similarity in any downstream method, holding for 2 to $n$ modality and providing more meaningful alignment with respect to previous similarity measures. The novel GRAM-based contrastive loss function enhances the alignment of multimodal models in the higher-dimensional embedding space, leading to new state-of-the-art performance in downstream tasks such as video-audio-text retrieval and audio-video classification. The project page, the code, and the pretrained models are available at https://ispamm.github.io/GRAM/.
This paper delves into the applications of generative artificial intelligence (GAI) in semantic communication (SemCom) and presents a thorough study. Three popular SemCom systems enabled by classical GAI models are first introduced, including variational autoencoders, generative adversarial networks, and diffusion models. For each system, the fundamental concept of the GAI model, the corresponding SemCom architecture, and the associated literature review of recent efforts are elucidated. Then, a novel generative SemCom system is proposed by incorporating the cutting-edge GAI technology-large language models (LLMs). This system features two LLM-based AI agents at both the transmitter and receiver, serving as "brains" to enable powerful information understanding and content regeneration capabilities, respectively. This innovative design allows the receiver to directly generate the desired content, instead of recovering the bit stream, based on the coded semantic information conveyed by the transmitter. Therefore, it shifts the communication mindset from "information recovery" to "information regeneration" and thus ushers in a new era of generative SemCom. A case study on point-to-point video retrieval is presented to demonstrate the superiority of the proposed generative SemCom system, showcasing a 99.98% reduction in communication overhead and a 53% improvement in retrieval accuracy compared to the traditional communication system. Furthermore, four typical application scenarios for generative SemCom are delineated, followed by a discussion of three open issues warranting future investigation. In a nutshell, this paper provides a holistic set of guidelines for applying GAI in SemCom, paving the way for the efficient implementation of generative SemCom in future wireless networks.
Current video retrieval systems, especially those used in competitions, primarily focus on querying individual keyframes or images rather than encoding an entire clip or video segment. However, queries often describe an action or event over a series of frames, not a specific image. This results in insufficient information when analyzing a single frame, leading to less accurate query results. Moreover, extracting embeddings solely from images (keyframes) does not provide enough information for models to encode higher-level, more abstract insights inferred from the video. These models tend to only describe the objects present in the frame, lacking a deeper understanding. In this work, we propose a system that integrates the latest methodologies, introducing a novel pipeline that extracts multimodal data, and incorporate information from multiple frames within a video, enabling the model to abstract higher-level information that captures latent meanings, focusing on what can be inferred from the video clip, rather than just focusing on object detection in one single image.
Existing large video-language models (LVLMs) struggle to comprehend long videos correctly due to limited context. To address this problem, fine-tuning long-context LVLMs and employing GPT-based agents have emerged as promising solutions. However, fine-tuning LVLMs would require extensive high-quality data and substantial GPU resources, while GPT-based agents would rely on proprietary models (e.g., GPT-4o). In this paper, we propose Video Retrieval-Augmented Generation (Video-RAG), a training-free and cost-effective pipeline that employs visually-aligned auxiliary texts to help facilitate cross-modality alignment while providing additional information beyond the visual content. Specifically, we leverage open-source external tools to extract visually-aligned information from pure video data (e.g., audio, optical character, and object detection), and incorporate the extracted information into an existing LVLM as auxiliary texts, alongside video frames and queries, in a plug-and-play manner. Our Video-RAG offers several key advantages: (i) lightweight with low computing overhead due to single-turn retrieval; (ii) easy implementation and compatibility with any LVLM; and (iii) significant, consistent performance gains across long video understanding benchmarks, including Video-MME, MLVU, and LongVideoBench. Notably, our model demonstrates superior performance over proprietary models like Gemini-1.5-Pro and GPT-4o when utilized with a 72B model.
Contextual advertising serves ads that are aligned to the content that the user is viewing. The rapid growth of video content on social platforms and streaming services, along with privacy concerns, has increased the need for contextual advertising. Placing the right ad in the right context creates a seamless and pleasant ad viewing experience, resulting in higher audience engagement and, ultimately, better ad monetization. From a technology standpoint, effective contextual advertising requires a video retrieval system capable of understanding complex video content at a very granular level. Current text-to-video retrieval models based on joint multimodal training demand large datasets and computational resources, limiting their practicality and lacking the key functionalities required for ad ecosystem integration. We introduce ContextIQ, a multimodal expert-based video retrieval system designed specifically for contextual advertising. ContextIQ utilizes modality-specific experts-video, audio, transcript (captions), and metadata such as objects, actions, emotion, etc.-to create semantically rich video representations. We show that our system, without joint training, achieves better or comparable results to state-of-the-art models and commercial solutions on multiple text-to-video retrieval benchmarks. Our ablation studies highlight the benefits of leveraging multiple modalities for enhanced video retrieval accuracy instead of using a vision-language model alone. Furthermore, we show how video retrieval systems such as ContextIQ can be used for contextual advertising in an ad ecosystem while also addressing concerns related to brand safety and filtering inappropriate content.
We introduce a goal-oriented conversational AI system enhanced with American Sign Language (ASL) instructions, presenting the first implementation of such a system on a worldwide multimodal conversational AI platform. Accessible through a touch-based interface, our system receives input from users and seamlessly generates ASL instructions by leveraging retrieval methods and cognitively based gloss translations. Central to our design is a sign translation module powered by Large Language Models, alongside a token-based video retrieval system for delivering instructional content from recipes and wikiHow guides. Our development process is deeply rooted in a commitment to community engagement, incorporating insights from the Deaf and Hard-of-Hearing community, as well as experts in cognitive and ASL learning sciences. The effectiveness of our signing instructions is validated by user feedback, achieving ratings on par with those of the system in its non-signing variant. Additionally, our system demonstrates exceptional performance in retrieval accuracy and text-generation quality, measured by metrics such as BERTScore. We have made our codebase and datasets publicly accessible at https://github.com/Merterm/signed-dialogue, and a demo of our signed instruction video retrieval system is available at https://huggingface.co/spaces/merterm/signed-instructions.
Efficiently retrieving and synthesizing information from large-scale multimodal collections has become a critical challenge. However, existing video retrieval datasets suffer from scope limitations, primarily focusing on matching descriptive but vague queries with small collections of professionally edited, English-centric videos. To address this gap, we introduce $\textbf{MultiVENT 2.0}$, a large-scale, multilingual event-centric video retrieval benchmark featuring a collection of more than 218,000 news videos and 3,906 queries targeting specific world events. These queries specifically target information found in the visual content, audio, embedded text, and text metadata of the videos, requiring systems leverage all these sources to succeed at the task. Preliminary results show that state-of-the-art vision-language models struggle significantly with this task, and while alternative approaches show promise, they are still insufficient to adequately address this problem. These findings underscore the need for more robust multimodal retrieval systems, as effective video retrieval is a crucial step towards multimodal content understanding and generation tasks.