Abstract: Conversational emotion recognition (CER) is an important research topic in human-computer interactions. Although recent advancements in transformer-based cross-modal fusion methods have ...
@inproceedings{Tang2024DRMF, title={DRMF: Degradation-Robust Multi-Modal Image Fusion via Composable Diffusion Prior}, author={Tang, Linfeng and Deng, Yuxin and Yi, Xunpeng and Yan, Qinglong and Yuan, ...
Abstract: A critical challenge for multi-modal Object Re-Identification (ReID) is the effective aggregation of complementary information to mitigate illumination issues. State-of-the-art methods ...
Towards open-ended Video Anomaly Detection (VAD), existing methods often exhibit biased detection when faced with challenging or unseen events and lack interpretability. To address these drawbacks, we ...