A computing device, such as any of the computing devices described for FIGS. 1 and 2, may be configured to generate any of the prediction models described herein. For example, FIG. 7 illustrates a flow diagram of an exemplary method 700 for generating an image-based prediction model that uses Distification. The method begins (block 702) where a computing device obtains a set of three dimensional (3D) images from a 3D image data source (block 704). The data source can include, for example, any of the computing devices, such as cameras, computers, servers, or remote computing devices as describe for FIGS. 1 and 2. Each 3D image in the set can be associated with 3D point cloud data as described herein. The 3D point cloud data can either be computed before the image is obtained or afterwards.
At block 706, the computing device can then apply Distification to the 3D point cloud of the respective images, as described herein for FIGS. 3-5. The Distification process can generate output feature vectors associated with the 3D images. In certain embodiments, an output feature vector may be generated for each 3D image. In other embodiments, an output feature vector may be generated for several 3D images, where each of the several 3D images would correspond to a single output feature vector.
