Fine-grained video action recognition can be formulated as a video-text matching problem. Previous approaches primarily rely on global video semantics to consolidate video embeddings, often leading to misaligned video-text pairs due to inaccurate atomic-level action understanding. This inaccuracy arises due to i) videos with distinct global semantics may share similar atomic actions or visual appearances, and ii) atomic actions can be momentary, gradual, or not directly aligned with overarching video semantics. Inspired by storyboarding, where a script is segmented into individual shots, we propose a multi-granularity framework, SFAR. SFAR generates fine-grained descriptions of common atomic actions for each global semantic using a large language model. Unlike existing works that refine global semantics with auxiliary video frames, SFAR introduces a filtering metric to ensure correspondence between the descriptions and the global semantics, eliminating the need for direct video involvement and thereby enabling more nuanced recognition of subtle actions. By leveraging both global semantics and fine-grained descriptions, our SFAR effectively identifies prominent frames within videos, thereby improving the accuracy of embedding aggregation. Extensive experiments on various video action recognition datasets demonstrate the competitive performance of our SFAR in supervised, few-shot, and zero-shot settings.

We introduce DanmakuTPPBench, a comprehensive benchmark designed to advance multi-modal Temporal Point Process (TPP) modeling in the era of Large Language Models (LLMs). While TPPs have been widely studied for modeling temporal event sequences, existing datasets are predominantly unimodal, hindering progress in models that require joint reasoning over temporal, textual, and visual information. To address this gap, DanmakuTPPBench comprises two complementary components: (1) DanmakuTPP-Events, a novel dataset derived from the Bilibili video platform, where user-generated bullet comments (Danmaku) naturally form multi-modal events annotated with precise timestamps, rich textual content, and corresponding video frames; (2) DanmakuTPP-QA, a challenging question-answering dataset constructed via a novel multi-agent pipeline powered by state-of-the-art LLMs and multi-modal LLMs (MLLMs), targeting complex temporal-textual-visual reasoning. We conduct extensive evaluations using both classical TPP models and recent MLLMs, revealing significant performance gaps and limitations in current methods' ability to model multi-modal event dynamics. Our benchmark establishes strong baselines and calls for further integration of TPP modeling into the multi-modal language modeling landscape.

When instructed to place a floor lamp next to an armchair, humans can visually ground it in the scene, estimating its base diameter and height, imagining its precise alignment with the armchair, and judging whether it fits naturally within the 3D environment. Humans can naturally perceive, reason about, and localize expressions to "anywhere" in 3D scenes. Yet can today's 3D vision-language models ground free-form referring expressions to precise positions and dimensions in a 3D scene, especially when those expressions refer to regions beyond objects? Existing 3D visual grounding models, pretrained on large 3D scene datasets, excel at aligning expressions to objects in a scene [7, 58, 2, 63, 61, 26]. However, these models remain constrained to object-level alignment, with limited attention paid to the broader spatial regions beyond objects.

Multimodal large language models (MLLMs) demonstrates remarkable capabilities in handling complex multimodal tasks and are increasingly adopted in video understanding applications. However, their rapid advancement raises serious data privacy concerns, particularly given the potential inclusion of sensitive video content, such as personal recordings and surveillance footage, in their training datasets. Determining improperly used videos during training remains a critical and unresolved challenge. Despite considerable progress on membership inference attacks (MIAs) for text and image data in MLLMs, existing methods fail to generalize effectively to the video domain. These methods suffer from poor scalability as more frames are sampled and generally achieve negligible true positive rates at low false positive rates (TPR@Low FPR), mainly due to their failure to capture the inherent temporal variations of video frames and to account for model behavior differences as the number of frames varies.

Potential negative societal impacts Although our work improves the performance of text-video retrieval, but may reduce the difficulty of cross-modal retrieval of sensitive information on the network. It may raise challenges to protecting information security. Limitations of our work Iterative approaches are sensitive to initialization and parameters such as the dimensions and the number of subspaces. In our work, although we use the L2 normalization operation to limit the value range of the parameters, the EM algorithm [3] may still converge to bad results. At the same time, the selection of the number of subspaces also has a relatively significant impact on the model effect.

Video Semantic Segmentation (VSS) involves assigning a semantic label to each pixel in a video sequence. Prior work in this field has demonstrated promising results by extending image semantic segmentation models to exploit temporal relationships across video frames; however, these approaches often incur significant computational costs. In this paper, we propose an efficient mask propagation framework for VSS, called MPVSS. Our approach first employs a strong querybased image segmentor on sparse key frames to generate accurate binary masks and class predictions. We then design a flow estimation module utilizing the learned queries to generate a set of segment-aware flow maps, each associated with a mask prediction from the key frame.

We introduce motion graph, a novel approach to address the video prediction problem, i.e., predicting future video frames from limited past data. The motion graph transforms patches of video frames into interconnected graph nodes, to comprehensively describe the spatial-temporal relationships among them. This representation overcomes the limitations of existing motion representations such as image differences, optical flow, and motion matrix that either fall short in capturing complex motion patterns or suffer from excessive memory consumption. We further present a video prediction pipeline empowered by motion graph, exhibiting substantial performance improvements and cost reductions. Extensive experiments on various datasets, including UCF Sports, KITTI and Cityscapes, highlight the strong representative ability of motion graph. Especially on UCF Sports, our method matches and outperforms the SOTA methods with a significant reduction in model size by 78% and a substantial decrease in GPU memory utilization by 47%.

Our goal is to predict future video frames given a sequence of input frames. Despite large amounts of video data, this remains a challenging task because of the high-dimensionality of video frames. We address this challenge by proposing the Decompositional Disentangled Predictive Auto-Encoder (DDPAE), a framework that combines structured probabilistic models and deep networks to automatically (i) decompose the high-dimensional video that we aim to predict into components, and (ii) disentangle each component to have low-dimensional temporal dynamics that are easier to predict. Crucially, with an appropriately specified generative model of video frames, our DDPAE is able to learn both the latent decomposition and disentanglement without explicit supervision. For the Moving MNIST dataset, we show that DDPAE is able to recover the underlying components (individual digits) and disentanglement (appearance and location) as we would intuitively do. We further demonstrate that DDPAE can be applied to the Bouncing Balls dataset involving complex interactions between multiple objects to predict the video frame directly from the pixels and recover physical states without explicit supervision.