Systems and method for textile fabric construction
Disclosed herein are systems and methods that allow an image of a fabric sample to be captured. The image may be analyzed to extract features, sometimes referred to as a stitch parameter, of the fabric's construction, such the use of one or more stitch types. Using the stitch parameters, a build specification can be selected from a plurality of build specification. Each of the plurality of build specifications may be associated with at least one of a plurality of known fabric constructions. The build specification may be exported and then used to construct a fabric similar to the fabric sample captured in the received image.
1. A method for training a machine learning model to recognize a fabric construction, the method comprising:
receiving a first subset of training images;
extracting stitch parameters including a stitch type, a courses per inch parameter, a wales per inch parameter and a yarn thickness for at least one strand of yarn from each of the first subset of training images;
mapping the stitch parameters from each of the first subset of training images to a variable of the machine learning model;
receiving a second subset of training images;
extracting the stitch parameters from each of the second subset of training images;
evaluating the machine learning model using the stitch parameters extracted from each of the second subset of training images;
receiving a confirmation that the machine learning model correctly identified a fabric construction for a majority of the second subset of training images; and
exporting the machine learning model in response to receiving the confirmation.
2. The method of claim 1 , wherein extracting the stitch parameters including performing an image analysis of the image to identify the stitch parameters.
3. The method of claim 1 , further comprising resizing the image of the fabric sample.
4. The method of claim 1 , further comprising converting a type of the image of the fabric sample from a first type to a second type.
5. The method of claim 1 , further comprising removing duplicate images from the first subset of training images.
6. The method of claim 1 , further comprising removing duplicate images from the second subset of training images.
7. The method of claim 1 , wherein receiving the first and second subsets of training images comprises:
receiving a plurality of images; and
dividing the plurality of images into the first and second subsets of training images.
8. The method of claim 1 , wherein receiving the first and second subset of training images comprises generating the first and second subset of training images.
9. A system for training a machine learning model to recognize a fabric construction, the system comprising:
at least one processor; and
at least one memory storing instructions that, when executed by the at least one processor, cause the at least one processor to perform actions comprising:
receiving a first subset of training imaging,
extracting stitch parameters including a stitch type, a courses per inch parameter, a wales per inch parameter and a yarn thickness for at least one strand of yarn from each of the first subset of training images,
mapping the stitch parameters from each of the first subset of training images to a variable of the machine learning model,
receiving a second subset of training images, extracting the stitch parameters from each of the second subset of training images,
evaluating the machine learning model using the stitch parameters extracted from each of the second subset of training images,
receiving a confirmation that the machine learning model correctly identified a fabric construction for a majority of the second subset of training images, and
exporting the machine learning model in response to receiving the confirmation.
10. The system of claim 9 , wherein extracting the stitch parameters including performing an image analysis of the image to identify the stitch parameters.
11. The system of claim 10 , wherein performing the image analysis includes image subtracting utilizing the image of the fabric sample and an image of a known fabric sample.
12. The system of claim 9 , wherein performing the image analysis includes extracting a vertical and horizontal dimension from the image of the fabric sample.
13. The system of claim 9 , wherein herein the actions further comprise resizing the image of the fabric sample.
14. The system of claim 9 , wherein the actions further comprise converting a type of the image of the fabric sample from a first type to a second type.
15. The system of claim 9 , wherein the actions further comprise removing duplicate images from the first subset of training images and the second subset of training images.
16. The system of claim 9 , wherein receiving the first and second subsets of training images comprises:
receiving a plurality of images; and
dividing the plurality of images into the first and second subsets of training images.
17. The system of claim 9 , wherein receiving the first and second subset of training images comprises generating the first and second subset of training images.