Masked Image Modeling for Efficient and Generalizable Anomaly Detection in Aerial Agricultural Images

A self-supervised learning approach using Masked Image Modeling can effectively detect a wide range of anomalies in aerial agricultural images without the need for extensive labeled data.