Anomaly Detection Across Multiple Farms Through Remote Sensing

In effective agricultural management, monitoring crop growth is crucial. Various factors, including climate, pests, diseases, and soil variations, can lead to abnormal yields. Detecting anomalies, where one farm’s growth deviates from its neighbors with similar crops and sowing dates, can help identify early signs of crop diseases. Remote sensing technologies, like satellites and drones, provide real-time crop growth information, enabling informed decisions for farmers. This paper presents a satellite-based anomaly detection technique. We extract NDVI values, sowing dates, and current dates for each farm, normalize the data, and employ the DBSCAN algorithm to detect anomalies. Our results demonstrate that DBSCAN outperforms other models with a silhouette score of 0.5369, improving crop yield prediction.