Abstract: The intelligent monitoring of ion−type rare earth mining operations is of great significance for ecological environment monitoring and the sustainable development of resources. With the rapid advancement of drone remote sensing, computer technology, and deep learning, high−resolution image data has provided new methods for detecting and extracting illegal mining activities in ion−type rare earth mining areas. Against this backdrop, a detection method for illegal mining in rare earth mining areas based on an improved YOLOv8s algorithm was proposed. This method incorporates a small−object detection layer, enhancing the accuracy of detecting illegal mining with small and concealed targets. Finally, the VoVGSCSP feature extraction module was introduced into the neck network to further optimize the feature transmission network, thereby improving the algorithm's detection performance and operational efficiency. Experimental results demonstrate that the proposed improved method significantly enhances the detection of illegal mining in rare earth mining areas, achieving an average precision and F1 score of 78.7% and 77%, respectively. Compared to the baseline algorithm, the improved YOLOv8s algorithm achieves a 2.8% increase in average detection precision and a 4% improvement in F1 score. When compared to various object detection algorithms, the proposed method exhibits clear advantages, surpassing Faster R−CNN by 19.22% and 31% in average precision and F1 score, respectively. Additionally, this algorithm enables rapid and precise detection of concealed illegal mining activities in complex natural environments of rare earth mining areas while effectively improving the identification and localization capabilities of illegal mining targets. This method can provide accurate and effective technical support for the intelligent monitoring of ion−type rare earth mining operations.