Visualization of ai-generated predictions from 3d data

A method, comprising:

• in an electronic apparatus that includes a memory for storing a three-dimensional object detection model:
• acquiring 3D data associated with a 3D environment that includes a 3D object;
• partitioning the 3D data into a set of data blocks;
• generating variations of the 3D data based on a modification of the set of data blocks;
• generating 3D object detection results corresponding to the generated variations based on an application of the 3D object detection model on the generated variations;
• training a regression model based on the 3D object detection results;
• determining a weight value that indicates a likelihood that a data block of the set of data blocks belongs to the 3D object,
• wherein the weight value is determined based on the trained regression model;
• and
• generating a first 3D heatmap visualization of the 3D object included in the 3D data based on the determined weight value for each data block of the set of data blocks.

A non-transitory computer-readable medium having stored thereon, computer-executable instructions that when executed by an electronic apparatus, causes the electronic apparatus to execute operations, the operations comprising:

• acquiring 3D data associated with a 3D environment that includes a 3D object;
• partitioning the 3D data into a set of data blocks;
• generating variations of the 3D data based on a modification of the set of data blocks;
• generating 3D object detection results corresponding to the generated variations based on an application of a 3D object detection model on the generated variations;
• training a regression model based on the 3D object detection results;
• determining a weight value that indicates a likelihood that a data block of the set of data blocks belongs to the 3D object,
• wherein the weight value is determined based on the trained regression model;
• and
• generating a first 3D heatmap visualization of the 3D object included in the 3D data based on the determined weight value for each data block of the set of data blocks.