Video understanding endeavors to generate descriptive texts by analyzing structured semantics from dynamic visual sequences, thus facilitating context-aware reasoning and interpretation. Recent advancements primarily rely on patch-level visual–textual alignment, bridging the gap between visual and textual modalities and therefore enabling more comprehensive reasoning. While promising, they often struggle to capture object-level semantics and temporal dependencies, resulting in limited interpretability and suboptimal compositional understanding. To address these issues, we propose a novel temporal relation framework for multi-modal video understanding, dubbed VideoU-MTR, which explicitly models object-level temporal relations to facilitate fine-grained and coherent representations of cross-frame object interactions. Specifically, we introduce a query-oriented frames identification mechanism that synergistically combines textual and visual attention, allowing the model to dynamically attend to semantically relevant video content across hierarchical levels while effectively filtering out irrelevant information. Furthermore, we employ an explicit temporal relation module to capture fine-grained temporal dependencies and inter-object dynamics by modeling object-centric sequences with time-aware attention and frame-level embeddings. Additionally, we propose a cross-modal alignment adapter that aligns temporally contextualized visual features with linguistic semantics at both object and frame levels. Extensive experiments on eight benchmarks across video question answering (VideoQA), long-term video understanding (LTVU), and video captioning (VideoCap) benchmarks demonstrate that VideoU-MTR achieves superior performance compared to state-of-the-art methods. Moreover, visualization analysis further validates the effectiveness of incorporating temporal information for enhancing video comprehension.

Can we accurately identify the true correspondences from multimodal datasets containing mismatched data pairs? Existing methods primarily emphasize the similarity matching between the representations of objects across modalities, potentially neglecting the crucial relation consistency within modalities that are particularly important for distinguishing the true and false correspondences. Such an omission often runs the risk of misidentifying negatives as positives, thus leading to unanticipated performance degradation. To address this problem, we propose a general Relation Consistency learning framework, namely ReCon, to accurately discriminate the true correspondences among the multimodal data and thus effectively mitigate the adverse impact caused by mismatches. Specifically, ReCon leverages a novel relation consistency learning to ensure the dual-alignment, respectively of, the cross-modal relation consistency between different modalities and the intra-modal relation consistency within modalities. Thanks to such dual constrains on relations, ReCon significantly enhances its effectiveness for true correspondence discrimination and therefore reliably filters out the mismatched pairs to mitigate the risks of wrong supervisions. Extensive experiments on three widely-used benchmark datasets, including Flickr30K, MS-COCO, and Conceptual Captions, are conducted to demonstrate the effectiveness and superiority of ReCon compared with other SOTAs. The code is available at: https://github.com/qxzha/ReCon.

Dynamic Facial Expression Recognition (DFER) facilitates the understanding of psychological intentions through non-verbal communication. Existing methods struggle to manage irrelevant information, such as background noise and redundant semantics, which impacts both efficiency and effectiveness. In this work, we propose a novel supervised temporal soft masked autoencoder network for DFER, namely AdaTosk, which integrates a parallel supervised classification branch with the self-supervised reconstruction branch. The self-supervised reconstruction branch applies random binary hard mask to generate diverse training samples, encouraging meaningful feature representations in visible tokens. Meanwhile the classification branch employs an adaptive temporal soft mask to flexibly mask visible tokens based on their temporal significance. Its two key components, respectively of, class-agnostic and class-semantic soft masks, serve to enhance critical expression moments and reduce semantic redundancy over time. Extensive experiments conducted on widely-used benchmarks demonstrate that our AdaTosk remarkably reduces computational costs compared with current state-of-the-art methods while still maintaining competitive performance.

The manual annotation of perfectly aligned labels for cross-modal retrieval (CMR) is incredibly labor-intensive. As an alternative, the collection of co-occurring data pairs from the Internet is a remarkably cost-effective way, but which, inevitably induces the Partially Mismatched Pairs (PMPs) and therefore significantly degrades the retrieval performance without particular treatment. Previous efforts often utilize the pair-wise similarity to filter out the mismatched pairs, and such operation is highly sensitive to mismatched or ambiguous data and thus leads to sub-optimal performance. To alleviate these concerns, we propose an efficient approach, termed UCPM, i.e., Uncertainty-guided Cross-modal retrieval with Partially Mismatched pairs, which can significantly reduce the adverse impact of mismatched data pairs. Specifically, a novel Uncertainty Guided Division (UGD) strategy is sophisticatedly designed to divide the corrupted training data into confident matched (clean), easily-identifiable mismatched (noisy) and hardly-determined hard subsets, and the derived uncertainty can simultaneously guide the informative pair learning while reducing the negative impact of potential mismatched pairs. Meanwhile, an effective Uncertainty Self-Correction (USC) mechanism is concurrently presented to accurately identify and rectify the fluctuated uncertainty during the training process, which further improves the stability and reliability of the estimated uncertainty. Besides, a Trusted Margin Loss (TML) is newly designed to enhance the discriminability between those hard pairs, by dynamically adjusting their soft margins to amplify the positive contributions of matched pairs while suppressing the negative impacts of mismatched pairs. Extensive experiments on three widely-used benchmark datasets, verify the effectiveness and reliability of UCPM compared with the existing SOTA approaches, and significantly improve the robustness in both synthetic and real-world PMPs. The code is available at: https://github.com/qxzha/UCPM.