Three-dimensional target detection method based on local features of key point cloud

The invention discloses a three-dimensional target detection method based on key point cloud local features. The method comprises the following steps: acquiring to-be-detected point cloud data at the current moment; inputting the obtained point cloud data into a trained multi-dimensional feature fusion feature coding network to obtain a target detection result, performing key point sampling and feature coding on the point cloud data in a training sample by the multi-dimensional feature fusion feature coding network to obtain key point point cloud local features, and specifically obtaining N pieces of original point cloud data; sampling N '/3 points from the N points as key points through a space distance farthest point sampling method; sampling N '/3 points from the N points through a feature distance farthest point sampling method; according to the method, the extracted local features of the object are more accurate, semantic information of the object can be better expressed, and subsequent object detection