Utilizing soft classifications to select input parameters for segmentation algorithms and identify segments of three-dimensional digital models
The present disclosure includes methods and systems for identifying and manipulating a segment of a three-dimensional digital model based on soft classification of the three-dimensional digital model. In particular, one or more embodiments of the disclosed systems and methods identify a soft classification of a digital model and utilize the soft classification to tune segmentation algorithms. For example, the disclosed systems and methods can utilize a soft classification to select a segmentation algorithm from a plurality of segmentation algorithms, to combine segmentation parameters from a plurality of segmentation algorithms, and/or to identify input parameters for a segmentation algorithm. The disclosed systems and methods can utilize the tuned segmentation algorithms to accurately and efficiently identify a segment of a three-dimensional digital model.
1. A computer-implemented method, comprising:
determining a soft classification of a three-dimensional digital model utilizing a classification algorithm, the soft classification comprising a categorization of the three-dimensional digital model into one or more model categories of a plurality of predefined model categories;
in response to determining the soft classification of the three-dimensional digital model, selecting input parameters for a segmentation algorithm utilizing the soft classification of the three-dimensional digital model; and
identifying a segment of the three-dimensional digital model utilizing the segmentation algorithm and the input parameters.
2. The computer-implemented method of claim 1 , further comprising:
determining correspondences between soft classifications and input parameter sets for the segmentation algorithm; and
selecting the input parameters for the segmentation algorithm based on the correspondences between the soft classifications and the input parameter sets.
3. The computer-implemented method of claim 2 , further comprising determining the correspondences by:
identifying a first accuracy metric for a first input parameter set for the soft classification; and
identifying a second accuracy metric for a second input parameter set for the soft classification.
4. The computer-implemented method of claim 3 , further comprising determining a correspondence between the soft classification and the first input parameter set by comparing the first accuracy metric and the second accuracy metric.
5. The computer-implemented method of claim 1 , further comprising selecting the input parameters by determining at least one of: a sensitivity metric, a segmentation threshold, or a surface normal significance metric.
6. The computer-implemented method of claim 1 , further comprising:
determining an additional soft classification of an additional three-dimensional digital model utilizing the classification algorithm; and
selecting additional input parameters for the segmentation algorithm utilizing the additional soft classification of the three-dimensional digital model, wherein the additional input parameters are different than the input parameters.
7. The computer-implemented method of claim 1 , wherein determining the soft classification of the three-dimensional digital model utilizing the classification algorithm comprises utilizing a machine learning model to generate classification probabilities for a plurality of soft classification categories.
8. The computer-implemented method of claim 1 , wherein identifying the segment of the three-dimensional digital model utilizing the segmentation algorithm and the input parameters comprises:
utilizing the input parameters to determine segmentation parameters;
receiving a user selection of a portion of the three-dimensional digital model; and
determining the segment utilizing the user selection and the segmentation parameters.
9. A non-transitory computer readable medium storing instructions, that when executed by at least one processor, cause the at least one processor to perform operations comprising:
determining a soft classification of a three-dimensional digital model utilizing a classification algorithm, the soft classification comprising a categorization of the three-dimensional digital model into one or more model categories of a plurality of predefined model categories;
in response to determining the soft classification of the three-dimensional digital model, selecting input parameters for a segmentation algorithm utilizing the soft classification of the three-dimensional digital model; and
identifying a segment of the three-dimensional digital model utilizing the segmentation algorithm and the input parameters.
10. The non-transitory computer readable medium of claim 9 , further comprising instructions, that when executed by the at least one processor, cause the at least one processor to perform operations comprising:
identifying a first accuracy metric for a first input parameter set for the soft classification;
identifying a second accuracy metric for a second input parameter set for the soft classification; and
selecting the input parameters for the segmentation algorithm based on the first accuracy metric and the second accuracy metric.
11. The non-transitory computer readable medium of claim 10 , wherein identifying the first accuracy metric comprises:
generating a predicted segmentation of a training digital model corresponding to the soft classification utilizing the first input parameter set; and
comparing the predicted segmentation to a known segmentation of the training digital model.
12. The non-transitory computer readable medium of claim 9 , further comprising instructions, that when executed by the at least one processor, cause the at least one processor to perform operations comprising selecting the input parameters by determining a surface normal significance metric.
13. The non-transitory computer readable medium of claim 9 , further comprising instructions, that when executed by the at least one processor, cause the at least one processor to perform operations comprising:
determining an additional soft classification of an additional three-dimensional digital model utilizing the classification algorithm;
selecting additional input parameters for the segmentation algorithm utilizing the additional soft classification of the three-dimensional digital model; and
identifying an additional segment of the additional three-dimensional digital model utilizing the segmentation algorithm and the additional input parameters.
14. The non-transitory computer readable medium of claim 9 , wherein determining the soft classification of the three-dimensional digital model utilizing the classification algorithm comprises:
utilizing a neural network to generate classification probabilities for a plurality of soft classification categories; and
determining the soft classification based on the plurality of soft classification categories.
15. The non-transitory computer readable medium of claim 9 , wherein identifying the segment of the three-dimensional digital model utilizing the segmentation algorithm and the input parameters comprises:
utilizing the input parameters to determine segmentation parameters; and
determining the segment corresponding utilizing the segmentation parameters.
16. A system comprising:
one or more memory devices; and
one or more processors, coupled to the one or more memory devices, that cause the system to perform operations comprising:
determining a soft classification of a three-dimensional digital model utilizing a classification algorithm, the soft classification comprising a categorization of the three-dimensional digital model into one or more model categories of a plurality of predefined model categories;
in response to determining the soft classification of the three-dimensional digital model, selecting input parameters for a segmentation algorithm utilizing the soft classification of the three-dimensional digital model; and
identifying a segment of the three-dimensional digital model utilizing the segmentation algorithm and the input parameters.
17. The system of claim 16 , wherein the one or more processors further cause the system to perform operations comprising:
determining a plurality of accuracy metrics for a plurality of input parameter sets by generating predicted segmentations utilizing the plurality of input parameter sets, the segmentation algorithm, and digital models corresponding to the soft classification; and
selecting the input parameters for the segmentation algorithm based on the plurality of accuracy metrics.
18. The system of claim 16 , wherein identifying the segment of the three-dimensional digital model utilizing the segmentation algorithm and the input parameters comprises:
utilizing the input parameters to determine edge segmentation scores;
receiving a user selection of a portion of the three-dimensional digital model; and
determining the segment utilizing the edge segmentation scores and the user selection.
19. The system of claim 16 , wherein determining the soft classification of the three-dimensional digital model utilizing at least one of a mixture of experts algorithm, a neural network, a decision tree, or a naïve Bayes algorithm to generate the soft classification from the three-dimensional digital model.
20. The system of claim 16 , wherein selecting the input parameters comprises determining a segmentation threshold for the segmentation algorithm.