This survey provides a comprehensive overview of recent advancements in Visual Anomaly Detection (VAD), highlighting three key challenges: 1) scarcity of training data, 2) diversity of visual modalities, and 3) complexity of hierarchical anomalies. The article begins by defining the concept of VAD and its applications in various domains, including industrial defect inspection and medical image analysis. It then discusses the challenges faced in VAD, such as the difficulty in acquiring sufficient abnormal samples for training, the use of diverse imaging modalities, and the hierarchical nature of anomalies. The survey categorizes recent VAD methods based on sample number, data modality, and anomaly hierarchy, analyzing their strengths and limitations. It also explores future directions for VAD, including the development of generic, multimodal, and holistic approaches. The article emphasizes the importance of addressing these challenges to improve the reliability and effectiveness of VAD systems in real-world applications. Key contributions of the survey include a detailed analysis of existing methods, identification of research trends, and recommendations for future research. The survey concludes that further advancements in VAD are necessary to achieve robust and efficient anomaly detection across diverse scenarios.This survey provides a comprehensive overview of recent advancements in Visual Anomaly Detection (VAD), highlighting three key challenges: 1) scarcity of training data, 2) diversity of visual modalities, and 3) complexity of hierarchical anomalies. The article begins by defining the concept of VAD and its applications in various domains, including industrial defect inspection and medical image analysis. It then discusses the challenges faced in VAD, such as the difficulty in acquiring sufficient abnormal samples for training, the use of diverse imaging modalities, and the hierarchical nature of anomalies. The survey categorizes recent VAD methods based on sample number, data modality, and anomaly hierarchy, analyzing their strengths and limitations. It also explores future directions for VAD, including the development of generic, multimodal, and holistic approaches. The article emphasizes the importance of addressing these challenges to improve the reliability and effectiveness of VAD systems in real-world applications. Key contributions of the survey include a detailed analysis of existing methods, identification of research trends, and recommendations for future research. The survey concludes that further advancements in VAD are necessary to achieve robust and efficient anomaly detection across diverse scenarios.