Method for accelerating three-dimensional object segmentation with point cloud simplifications
Aspects of the subject disclosure may include, for example, a device that has a processing system including a processor; and a memory that stores executable instructions that, when executed by the processing system, facilitate performance of operations, including downsampling a full point cloud to obtain a downsampled point cloud, wherein the downsampling reduces a data size of the full point cloud; and using a machine-learning model to assign labels for segmentation and object identification to points in the downsampled point cloud, wherein the machine-learning model is trained on the full point cloud. Other embodiments are disclosed.
1 . A device, comprising:
a processing system including a processor; and
a memory that stores executable instructions that, when executed by the processing system, facilitate performance of operations, the operations comprising:
selecting an accuracy requirement for segmentation;
selecting an optimization goal based on the accuracy requirement;
selecting parameters for a machine-learning model based on the optimization goal, resulting in a parameterized model;
training the parameterized model on at least one full point cloud, resulting in a trained model;
selecting, based on the optimization goal, a downsampling technique, an amount of data reduction, or a combination thereof;
downsampling, based on the downsampling technique, the amount of data reduction, or the combination thereof, a point cloud representation of an object to generate a downsampled point cloud; and
segmenting, using the trained model, the downsampled point cloud to assign labels to points of the object.
2 . The device of claim 1 , wherein the downsampled point cloud is based on a use of grid simplification.
3 . The device of claim 1 , wherein the downsampled point cloud comprises 20 to 60 percent of points in the point cloud representation of the object.
4 . The device of claim 1 , wherein the downsampled point cloud comprises 63 to 87 percent of points in the point cloud representation of the object.
5 . The device of claim 1 , wherein the selecting of the downsampling technique, the amount of data reduction, or the combination thereof comprises selecting the downsampling technique.
6 . The device of claim 1 , wherein the selecting of the downsampling technique, the amount of data reduction, or the combination thereof comprises selecting the amount of data reduction.
7 . The device of claim 1 , wherein the processing system comprises a plurality of processors operating in a distributed computing environment, the plurality of processors including the processor.
8 . The device of claim 1 , wherein the labels are associated with a brand.
9 . The device of claim 8 , wherein the labels are used for targeted advertising in connection with the brand.
10 . A non-transitory machine-readable medium, comprising executable instructions that, when executed by a processing system including a processor, facilitate performance of operations, the operations comprising:
selecting an optimization goal based on an accuracy requirement;
selecting parameters for a model based on the optimization goal, resulting in a parameterized model;
training the parameterized model on a plurality of full point clouds, resulting in a trained model;
selecting, based on the optimization goal, a downsampling technique, an amount of data reduction, or a combination thereof;
downsampling, based on the downsampling technique, the amount of data reduction, or the combination thereof, a point cloud representation of an object to generate a downsampled point cloud; and
segmenting, using the trained model, the downsampled point cloud to assign a label to a point of the object.
11 . The non-transitory machine-readable medium of claim 10 , wherein the segmenting of the downsampled point cloud results in an assignment of a second label to a second point of the object.
12 . The non-transitory machine-readable medium of claim 10 , wherein the selecting of the downsampling technique, the amount of data reduction, or the combination thereof comprises selecting the downsampling technique and a downsampling reduction percentage to reduce memory usage while meeting a required segmentation accuracy.
13 . The non-transitory machine-readable medium of claim 10 , wherein the selecting of the downsampling technique, the amount of data reduction, or the combination thereof comprises selecting the downsampling technique and a downsampling reduction percentage to reduce computational overhead while meeting a required segmentation accuracy.
14 . The non-transitory machine-readable medium of claim 10 , wherein the parameters comprise a number of feature planes.
15 . The non-transitory machine-readable medium of claim 10 , wherein the parameters comprise a style of a submanifold block.
16 . A method, comprising:
selecting, by a processing system including a processor, an accuracy requirement for segmentation;
selecting, by the processing system, an optimization goal based on the accuracy requirement;
training a parameterized model on a full point cloud, resulting in a trained model;
selecting, based on the optimization goal, a downsampling technique, an amount of data reduction, or a combination thereof;
downsampling, based on the downsampling technique, the amount of data reduction, or the combination thereof, a point cloud representation of an object to generate a downsampled point cloud; and
segmenting, using the trained model, the downsampled point cloud to assign labels to points of the object.
17 . The method of claim 16 , wherein the selecting of the downsampling technique, the amount of data reduction, or the combination thereof comprises selecting the downsampling technique.
18 . The method of claim 17 , wherein the downsampling technique is applied in stages.
19 . The method of claim 18 , further comprising:
receiving, by the processing system, a user input that identifies regions of the downsampled point cloud that should receive more detail in a first stage of the stages; and
enhancing, by the processing system and based on the receiving of the user input, detail of the regions in a second stage of the stages.
20 . The method of claim 16 , comprising:
applying, by the processing system, parameter quantization to reduce a number of bits associated with the downsampled point cloud; and
applying, by the processing system, network pruning and sharing techniques to reduce a complexity of a network.