Seismographs are highly sensitive and broadband instrument designed for detecting and recording seismic waves. In recent years, seismographs have been widely applied in the field of traffic monitoring, demonstrating significant potential, particularly for monitoring low-altitude aircraft. Radar, the traditional tool for aircraft detection, provides accurate and real-time capabilities in locating aircraft position, speed, and flight direction. However, it has limitations in detecting low-altitude aircraft, flexibility, susceptibility to terrain masking, and accuracy under extreme weather conditions. To overcome these limitations, researchers have explored various alternative methods for low-altitude aircraft detection, including infrared, imaging, and acoustic detection. Nevertheless, these methods also face challenges related to cost, weather susceptibility, detection range, and environmental noise.To address these challenges, this paper proposes a real-time recognition method for low-altitude aircraft, based on seismographs and deep learning technology. Leveraging the high precision and anti-interference capability of seismographs and integrating deep learning techniques, the method employs a lightweight MobileNetV3 network model for training, achieving automated identification of low-altitude aircraft signals. Furthermore, the vibrational frequency characteristics of the observed aircraft signals, inverted over time, yielded key flight parameters. The accuracy of these derived parameters was validated through comparison with flight trajectory data.The proposed method exhibits real-time performance and high recognition accuracy for low-altitude aircraft identification. Simultaneously, benefiting from the portability and ease of deployment of seismographs, the method is adaptable to complex environmental application requirements. It offers a feasible technical solution for the field of air traffic monitoring.