传统杂物检测方法在脱泥筛中应用时,杂物的位置和运动轨迹随着筛机震动难以预测,且视野受到激振器等设备的遮挡,导致目标杂物在盲点区域内发生轨迹偏移。针对以上问题,提出了一种多相机重叠视域下的筛上杂物检测与跨镜追踪方法,利用两台相机协同工作。其中,第一台相机使用YOLOv5对杂物做目标检测,确定杂物位置并将结果映射到第二台相机的视域内。第二台相机基于SORT算法对初始帧初始坐标进行追踪,实时监测和更新杂物的轨迹。该方法在提高检测准确性和追踪效率的同时,降低了计算复杂性。

When the process traditional impurity detection methods is applied to desliming screens, the positions and trajectories of impurities on desliming screens are difficult to predict, and the field of view can be obstructed by equipment like exciters, leading to trajectory deviations of target impurities within blind spots. Aiming at the above problems, we propose a method for impurity detection on desliming screens and cross-camera tracking in an overlapping multi-camera field of view. It involves the collaborative work of two cameras. The first camera employs the YOLOv5 algorithm for impurity target detection, determining impurity positions and mapping the results into the field of view of the second camera. The second camera, using the SORT algorithm, tracks the initial coordinates from the first frame, continuously monitoring and updating the trajectories of impurities. This method not only enhances detection accuracy and tracking efficiency but also reduces computational complexity.