Supervised Anomaly Detection

In the field of computer vision, supervised anomaly detection is a critical task aimed at training models to identify anomalies in data using a limited number of abnormal samples and a large number of normal samples. The goal of this task is to enhance the model's sensitivity and accuracy in detecting anomalous data, addressing the issue of natural imbalance learning due to uneven data distribution. This makes it particularly valuable in applications such as industrial monitoring, medical diagnosis, and security protection.