Method and system to augment images and labels to be compatible with various machine learning models
A method includes obtaining an authentic image of an assembly and a boundary label provided with the authentic image. The boundary label is associated with a selected region of the authentic image depicting a selected object. The method includes generating an augmented image based on the authentic image and an augmentation model employing one or more augmentation parameters, defining a model boundary label on a blank image at a region that correlates with the selected region of the authentic image, generating an augmented blank image based on the one or more augmentation parameters employed for the augmented image, identifying, as an augmented boundary label associated with augmented image, the model boundary label in the augmented blank image, and outputting an augmented image data, wherein the augmented image data incudes data indicative of the augmented image and of the augmented boundary label.
1 . A method comprising:
obtaining an authentic image of an assembly and a boundary label provided with the authentic image, wherein the boundary label is associated with a selected region of the authentic image depicting a selected object;
generating an augmented image based on the authentic image and an augmentation model employing one or more augmentation parameters;
defining a model boundary label on a blank image at a region that correlates with the selected region of the authentic image;
generating an augmented blank image based on the one or more augmentation parameters employed for the augmented image;
identifying, as an augmented boundary label associated with the augmented image, the model boundary label as the augmented boundary label associated with the augmented image in the augmented blank image;
outputting an augmented image data, wherein the augmented image data incudes data indicative of the augmented image and of the model boundary label as the augmented boundary label; and
detecting defects in the assembly or identifying objects in an assembly process by training a machine learning model based on the augmented image data and the augmented boundary label.
2 . The method of claim 1 further comprising:
defining a locked boundary label based on positional data of the model boundary label, wherein:
the locked boundary label is outputted as the augmented boundary label, and
dimensions of the locked boundary label is defined to be unrestricted to dimensions of the boundary label of the authentic image.
3 . The method of claim 2 , wherein:
the locked boundary label is defined by a first boundary axis and a second boundary axis orthogonal to the first boundary axis,
the first boundary axis and the second boundary axis are parallel to a first model axis and a second model axis, respectively, and
the first model axis and the second model axis are orthogonal to each other and define a two-dimensional coordinate plane of the blank image.
4 . The method of claim 3 , wherein defining the locked boundary label further includes setting coordinates of the locked boundary label based on a maximum coordinate value and a minimum coordinate value of the model boundary label along each of the first model axis and the second model axis.
5 . The method of claim 2 further comprising determining whether a locked boundary label is desired, wherein, the defining the locked boundary label is provided in response to determining the locked boundary label is desired.
6 . The method of claim 1 , wherein the augmented blank image is defined by a plurality of pixels having a black pixel value.
7 . The method of claim 1 , wherein dimensions of the model boundary label is restricted to dimensions of the boundary label associated with the authentic image.
8 . The method of claim 1 , wherein an orientation of the model boundary label is unrestricted to an orientation of the boundary label associated with the authentic image.
9 . The method of claim 1 , wherein the identifying, as the augmented boundary label associated with the augmented image, the model boundary label in the augmented blank image using a contour identification software application.
10 . The method of claim 1 , wherein the augmentation parameters includes a rotation parameter, a vertical translation parameter, a horizontal translation parameter, or a combination thereof.
11 . A method comprising:
obtaining an authentic image of an assembly and a boundary label provided with the authentic image, wherein the boundary label is associated with a selected region of the authentic image depicting a selected object;
generating an augmented image based on the authentic image and an augmentation model employing one or more augmentation parameters;
determining whether the one or more augmentation parameters includes an orientation modification parameter;
outputting, in response to the one or more augmentation parameter including the orientation modification parameter, a first augmented image data, wherein the first augmented image data includes data indicative of the augmented image and of the boundary label; and
performing a boundary label augmentation routine, in response to the one or more augmentation parameter including the orientation modification parameter, wherein the boundary label augmentation routine comprises:
defining a model boundary label on a blank image at a region that correlates with the selected region of the authentic image;
generating an augmented blank image based on the one or more augmentation parameters employed for the augmented image;
identifying, as an augmented boundary label associated with the augmented image, the model boundary label as the augmented boundary label associated with the augmented image in the augmented blank image;
outputting a second augmented image data, wherein the second augmented image data includes data indicative of the augmented image and of the model boundary label as the augmented boundary label; and
detecting defects in the assembly or identifying objects in an assembly process by training a machine learning model based on the second augmented image data and the augmented boundary label.
12 . The method of claim 11 , wherein the augmentation parameters further includes a contrast parameter, a brightness parameter, a sharpness parameter, or a combination thereof.
13 . The method of claim 11 , wherein the boundary label augmentation routine further comprises:
defining a locked boundary label based on positional data of the model boundary label, wherein:
the locked boundary label is outputted as the augmented boundary label, and
dimensions of the locked boundary label is defined to be unrestricted to dimensions of the boundary label of the authentic image.
14 . The method of claim 13 , wherein:
the locked boundary label is defined by a first boundary axis and a second boundary orthogonal to the first boundary axis,
the first boundary axis and the second boundary axis are parallel to a first model axis and a second model axis, respectively, and
the first model axis and the second model axis are orthogonal to each other and define a two-dimensional coordinate plane of the blank image.
15 . The method of claim 14 , wherein defining the locked boundary label further includes setting coordinates of the locked boundary label based on a maximum coordinate value and a minimum coordinate value of the model boundary label along each of the first model axis and the second model axis.
16 . The method of claim 13 , wherein the boundary label augmentation routine further comprises determining whether a locked boundary label is desired, wherein, the defining the locked boundary label is provided in response to determining the locked boundary label is desired.
17 . The method of claim 11 , wherein the orientation modification parameter includes a rotation parameter, a vertical translation parameter, a horizontal translation parameter, or a combination thereof.
18 . The method of claim 11 , wherein dimensions of the model boundary label is restricted to dimensions of the boundary label associated with the authentic image.
19 . The method of claim 11 , wherein an orientation of the model boundary label is unrestricted to an orientation of the boundary label associated with the authentic image.
20 . The method of claim 11 , wherein the identifying, as the augmented boundary label associated with the augmented image, the model boundary label in the augmented blank image using a contour identification software application.