Earthquake Intensity Prediction Based on Visual Target Tracking

Earthquake intensity represents the strength of ground shaking and its impact caused by an earthquake. Accurately and rapidly predicting earthquake intensity is crucial for providing timely and precise alerts to potential users during earthquake early warning processes. Traditional earthquake intensity analysis relies on conventional seismic measuring in-struments and optical seismometers. Although widely used and technically mature, these methods are limited by slow response times and high equipment costs. This paper proposes a visual earthquake intensity prediction method based on an alternating detection-tracking approach. The method employs object tracking to obtain the motion trajectory of objects while integrating object detection algorithms. When the tracking model detects that the target has disappeared or the tracking becomes unreliable, the detection model is invoked to re- identify the target, enhancing responsiveness to changes in the target's state. The object's displacement, velocity, and acceleration are calculated from its motion trajectory, and the earthquake intensity is subsequently derived using the earthquake intensity calculation formula. The proposed algorithm is applied to earthquake intensity predictions for the Wenchuan, Maerkang, and Wushi earthquakes. Comparative experiments with multiple object tracking methods validate the effectiveness of the proposed approach.