The deep learning-based anomaly detection methods using visual sensors generally rely on a single modality or variants as raw signal inputs, which severely limits expressiveness and adaptability. The ...

To address the aforementioned issues, we propose a multimodal visual-language prompt network (MVLPNet), which employs a collaborative optimization strategy for visual–textual features to tackle the ...