Tower crane collision object early warning method based on point cloud segmentation and recognition

The invention provides a tower crane collision object early warning method based on point cloud segmentation and recognition, which comprises the following steps: installing acquisition equipment at the front end of a cargo boom of a tower crane, and acquiring point cloud data in real time and preprocessing the point cloud data; performing point cloud segmentation on the point cloud data based on an European clustering algorithm; a tower crane collision object data set is collected, the constructed deep learning model is trained, and the trained deep learning model is stored; inputting the point cloud data subjected to the point cloud segmentation into a trained deep learning model, and identifying the type of the collision object; and according to the identified collision object type, two-stage early warning is set according to the collision object type, and tower crane collision object early warning based on point cloud segmentation and identification is completed. The collision objects around the tower cra