Systems and methods for 3D image distification
摘要
說明書
For example, in certain embodiments, a Distify method can take the unstructured data of an original 3D image, such as from a PLY file, as input and can generate a uniform output feature vector by first creating a uniform 2D matrix of points. After creating the matrix, the Distify method can determine the nearest points in the original 3D point cloud of the 3D image with respect to one or more of the 2D matrix points. In certain embodiments, the output feature vector can contain a z-value of the nearest 3D point for one or more of the 2D matrix points in the 2D matrix. In other embodiments, the output feature vector can contain a distance value based on the distance between a 2D matrix point to a 3D point in the 3D point cloud.
A predictive model may be trained using one or more of the output feature vectors containing the 2D and 3D point data and machine learning techniques. Once the model is trained, future 2D and 3D point data may be used as input to the model so that the model can be used to make predictions. Such predictions can include, for example, determining or classifying a driver's behavior as described herein.
