RESUMO
Weakly supervised video anomaly detection aims to locate abnormal activities in untrimmed videos without the need for frame-level supervision. Prior work has utilized graph convolution networks or self-attention mechanisms alongside multiple instance learning (MIL)-based classification loss to model temporal relations and learn discriminative features. However, these approaches are limited in two aspects: 1) Multi-branch parallel architectures, while capturing multi-scale temporal dependencies, inevitably lead to increased parameter and computational costs. 2) The binarized MIL constraint only ensures the interclass separability while neglecting the fine-grained discriminability within anomalous classes. To this end, we introduce a novel WS-VAD framework that focuses on efficient temporal modeling and anomaly innerclass discriminability. We first construct a Temporal Context Aggregation (TCA) module that simultaneously captures local-global dependencies by reusing an attention matrix along with adaptive context fusion. In addition, we propose a Prompt-Enhanced Learning (PEL) module that incorporates semantic priors using knowledge-based prompts to boost the discrimination of visual features while ensuring separability across anomaly subclasses. The proposed components have been validated through extensive experiments, which demonstrate superior performance on three challenging datasets, UCF-Crime, XD-Violence and ShanghaiTech, with fewer parameters and reduced computational effort. Notably, our method can significantly improve the detection accuracy for certain anomaly subclasses and reduced the false alarm rate. Our code is available at: https://github.com/yujiangpu20/PEL4VAD.
RESUMO
Introduction: Vascular smooth muscle cells (VSMCs) are the predominant cell type in the medial layer of the aorta, which plays a critical role in aortic diseases. Innate immunity is the main driving force for cardiovascular diseases. Methods: To determine the roles of innate immunity in VSMC and aortic pathologies, we performed transcriptome analyses on aortas from ApoE-/- angiotensin II (Ang II)-induced aortic aneurysm (AAA) time course, and ApoE-/- atherosclerosis time course, as well as VSMCs stimulated with danger-associated molecular patterns (DAMPs). Results: We made significant findings: 1) 95% and 45% of the upregulated innate immune pathways (UIIPs, based on data of 1226 innate immune genes) in ApoE-/- Ang II-induced AAA at 7 days were different from that of 14 and 28 days, respectively; and AAA showed twin peaks of UIIPs with a major peak at 7 days and a minor peak at 28 days; 2) all the UIIPs in ApoE-/- atherosclerosis at 6 weeks were different from that of 32 and 78 weeks (two waves); 3) analyses of additional 12 lists of innate immune-related genes with 1325 cytokine and chemokine genes, 2022 plasma membrane protein genes, 373 clusters of differentiation (CD) marker genes, 280 nuclear membrane protein genes, 1425 nucleoli protein genes, 6750 nucleoplasm protein genes, 1496 transcription factors (TFs) including 15 pioneer TFs, 164 histone modification enzymes, 102 oxidative cell death genes, 68 necrotic cell death genes, and 47 efferocytosis genes confirmed two-wave inflammation in atherosclerosis and twin-peak inflammation in AAA; 4) DAMPs-stimulated VSMCs were innate immune cells as judged by the upregulation of innate immune genes and genes from 12 additional lists; 5) DAMPs-stimulated VSMCs increased trans-differentiation potential by upregulating not only some of 82 markers of 7 VSMC-plastic cell types, including fibroblast, osteogenic, myofibroblast, macrophage, adipocyte, foam cell, and mesenchymal cell, but also 18 new cell types (out of 79 human cell types with 8065 cell markers); 6) analysis of gene deficient transcriptomes indicated that the antioxidant transcription factor NRF2 suppresses, however, the other five inflammatory transcription factors and master regulators, including AHR, NF-KB, NOX (ROS enzyme), PERK, and SET7 promote the upregulation of twelve lists of innate immune genes in atherosclerosis, AAA, and DAMP-stimulated VSMCs; and 7) both SET7 and trained tolerance-promoting metabolite itaconate contributed to twin-peak upregulation of cytokines in AAA. Discussion: Our findings have provided novel insights on the roles of innate immune responses and nuclear stresses in the development of AAA, atherosclerosis, and VSMC immunology and provided novel therapeutic targets for treating those significant cardiovascular and cerebrovascular diseases.