Model training method and device

The invention discloses a model training method and device. The method comprises steps of obtaining point cloud containing position information of an actual bounding box as a training sample; inputting the training sample into a to-be-trained detection model; obtaining the position information and the prediction difference information of a prediction bounding box outputted by the to-be-trained detection model; wherein the prediction difference information is a prediction value of the difference between the position information of the prediction bounding box and the position information of theactual bounding box by the to-be-trained detection model; and determining the actual difference information according to the position information of the prediction bounding box and the position information of the actual bounding box, and training the to-be-trained detection model by taking minimization of the actual difference information and minimization of the prediction difference information as training targets. Throu