Collection of image data for use in training a machine-learning model
A method of collecting and conditioning image data of an item for use in training a machine-learning model to detect at least one defect can include capturing a first set of images by a first camera with each image of the first set of images having a first viewpoint of the item that is the same viewpoint as the other images in the first set of images and can be illuminated by various wavelengths of light including ultraviolet light, infrared light, and visible light. The method can further include examining one image of the first set of images to determine if the item contains at least one defect, identifying a first location of the at least one defect on the one image, designating the first location on the other images of the first set of images so that all images in the first set of images identify the first location.
1 . A method of collecting and conditioning image data of an item for use in training a machine-learning model to detect at least one defect, the method comprising:
capturing a first set of images by a first camera with each image of the first set of images having a first viewpoint of the item that is the same viewpoint as the other images in the first set of images, wherein the first set of images includes images capturing the item as illuminated by various wavelengths of light including ultraviolet light, infrared light, and visible light;
examining one image of the first set of images to determine if the item contains at least one defect;
identifying, in response to the one image showing the item contains at least one defect, a first location of the at least one defect on the one image;
designating the first location, which corresponds to the at least one defect, on the other images of the first set of images so that all images in the first set of images identify the first location; and
adding the first set of images to a corpus having the image data for use in training the machine-learning model.
2 . The method of claim 1 , further comprising:
examining each image of the first set of images to determine if the item contains at least one defect; and
labeling, in response to all images in the first set of images not showing at least one defect in the item, the first set of images as showing no defect.
3 . The method of claim 2 , further comprising:
using the image data to train the machine-learning model to identify defects in components similar to the item.
4 . The method of claim 3 , wherein the machine-learning model includes computer vision.
5 . The method of claim 1 , further comprising:
labeling the first set of images as showing at least one defect.
6 . The method of claim 1 , further comprising:
capturing a second set of images of the item by a second camera with the second set of images having second viewpoint of the item that is the same viewpoint as the other images in the second set of images, wherein the second set of images includes images capturing the item as illuminated by various wavelengths of light including ultraviolet light, infrared light, and visible light;
examining one image of the second set of images to determine if the item contains at least one defect;
designating, in response to the one image showing the item contains at least one defect, a second location of the at least one defect on each image of the second set of images; and
adding the second set of images to the image data for use in training the machine-learning model.
7 . The method of claim 6 , wherein the first camera and the second camera are the same so as to be in the same position, the method further comprising:
rotating the item from an orientation in the first viewpoint to a different orientation before capturing the second set of images having the second viewpoint.
8 . The method of claim 6 , wherein the second camera is in a different position relative to the item so as to capture the second set of images from the second viewpoint that is different from the first viewpoint.
9 . The method of claim 1 , further comprising:
shining ultraviolet light at the item by an ultraviolet light source; and
capturing a first image of the first set of images of the item with the first image showing the item as illuminated by ultraviolet light.
10 . The method of claim 9 , further comprising:
shining infrared light at the item by an infrared light source; and
capturing a second image of the first set of images of the item with the second image showing the item as illuminated by infrared light.
11 . The method of claim 10 , further comprising:
shining visible light at the item; and
capturing a third image of the first set of images of the item with the third image showing the item as illuminated by visible light.
12 . The method of claim 1 , wherein the item has an anodized metal surface.
13 . The method of claim 1 , wherein the step of designating the first location on the other images of the first set of images is performed by a computer processor.
14 . The method of claim 13 , wherein the step of designating the first location on the other images of the first set of images is performed automatically by the computer processor in response to the identification of the first location on one image of the first set of images.
15 . A system for collecting and conditioning image data of an item for use in training a machine-learning model to detect at least one defect, the system comprising:
a light source configured to shine light having different wavelengths onto the item, the light including ultraviolet light, infrared light, and visible light;
a camera configured to capture a first set of images having a first viewpoint of the item that is the same viewpoint for each image in the first set of images, wherein the first set of images includes images captured at various wavelengths of light including ultraviolet light, infrared light, and visible light;
a user interface in communication with the camera to receive at least one of the first set of images, the user interface enabling examination of one image of the first set of images to determine if the item contains at least one defect and identification, in response to the determination that the one image shows at least one defect, of a first location of the at least one defect on the one image;
a computer processor in communication with the user interface and configured to designate the first location on the other images of the first set of images so that all images in the first set of images identify the first location; and
a corpus having image data that includes the first set of images with the corpus being used in training the machine-learning model.
16 . The system of claim 15 , wherein the designation of the first location on the other images of the first set of images is performed automatically by the computer processor in response to the identification of the first location on one image of the first set of images.
17 . The system of claim 15 , wherein the computer processor labels the first set of images as showing at least one defect.
18 . The system of claim 15 , further comprising:
the machine-learning model is provided the corpus having the image data for use in training the machine-learning model to identify defects in components similar to the item.
19 . The system of claim 15 , wherein the computer processor, in response to the examination of all images of the first set of images resulting in a determination that no defects in the item are shown in the images, labels the first set of images as showing no defect.
20 . The system of claim 15 , wherein the light source and the camera are contained within a housing.