Systems and methods for determining defects in physical objects
In one embodiment, a method includes receiving, by a defect detector module, an image of a physical object and classifying, by the defect detector module, one or more first features from the image of the physical object into one or more first classifications using one or more machine learning algorithms. The method further includes analyzing, by the defect detector module, the one or more first classifications and determining, by the defect detector module, that the physical object comprises a defect based on analyzing the one or more first classifications.
1. A method, comprising:
receiving, by a defect detector module, an image of a physical object;
classifying, by the defect detector module, one or more first features from the image of the physical object into one or more first classifications using one or more machine learning algorithms;
analyzing, by the defect detector module, the one or more first classifications;
determining, by the defect detector module, that the physical object comprises a defect based on analyzing the one or more first classifications;
classifying, by the defect detector module, one or more second features from the image of the physical object into one or more second classifications using the one or more machine learning algorithms; and
cropping, by the defect detector module, the image to an area surrounding the one or more second features to create a cropped image;
wherein classifying the one or more first features from the image of the physical object into the one or more first classifications comprises classifying the one or more first features from the cropped image of the physical object into the one or more first classifications.
2. The method of claim 1 , further comprising:
determining, by the defect detector module, a location of the defect within the image;
determining, by the defect detector module, that the location of the defect within the image is part of the physical object; and
determining, by the defect detector module, a geographic location of the defect of the physical object based at least in part on the location of the defect within the image.
3. The method of claim 1 , further comprising labeling, by the defect detector module, the one or more first features of the image with one or more labels.
4. The method of claim 1 , further comprising training, by the defect detector module, the one or more machine learning algorithms to classify the one or more first features from the image by collecting sample data representative of the one or more first features.
5. The method of claim 1 , wherein:
the physical object is a rail joint;
the defect is a broken rail joint; and
the one or more first classifications comprise one or more of the following:
a bolt;
a break;
a hole; and
a discontinuity.
6. The method of claim 1 ,
wherein the one or more second classifications comprise at least one of the following: a bar, a discontinuity, and an end post.
7. The method of claim 1 , wherein the image of the physical object is captured by a component in motion relative to the physical object.
8. A system comprising one or more processors and a memory storing instructions that, when executed by the one or more processors, cause the one or more processors to perform operations comprising:
receiving, by a defect detector module, an image of a physical object;
classifying, by the defect detector module, one or more first features from the image of the physical object into one or more first classifications using one or more machine learning algorithms;
analyzing, by the defect detector module, the one or more first classifications;
determining, by the defect detector module, that the physical object comprises a defect based on analyzing the one or more first classifications;
classifying, by the defect detector module, one or more second features from the image of the physical object into one or more second classifications using the one or more machine learning algorithms, wherein the one or more second classifications comprise at least one of the following: a bar, a discontinuity, and an end post; and
cropping, by the defect detector module, the image to an area surrounding the one or more second features to create a cropped image;
wherein classifying the one or more first features from the image of the physical object into the one or more first classifications comprises classifying the one or more first features from the cropped image of the physical object into the one or more first classifications.
9. The system of claim 8 , the operations further comprising:
determining, by the defect detector module, a location of the defect within the image;
determining, by the defect detector module, that the location of the defect within the image is part of the physical object; and
determining, by the defect detector module, a geographic location of the defect of the physical object based at least in part on the location of the defect within the image.
10. The system of claim 8 , the operations further comprising labeling, by the defect detector module, the one or more first features of the image with one or more labels.
11. The system of claim 8 , the operations further comprising training, by the defect detector module, the one or more machine learning algorithms to classify the one or more first features from the image by collecting sample data representative of the one or more first features.
12. The system of claim 8 , wherein:
the physical object is a rail joint;
the defect is a broken rail joint; and
the one or more first classifications comprise one or more of the following:
a bolt;
a break;
a hole; and
a discontinuity.
13. The system of claim 8 ,
wherein the one or more second classifications comprise at least one of the following: a bar, a discontinuity, and an end post.
14. The system of claim 8 , wherein the image of the physical object is captured by a component in motion relative to the physical object.
15. One or more non-transitory computer-readable storage media embodying instructions that, when executed by a processor, cause the processor to perform operations comprising:
receiving, by a defect detector module, an image of a physical object;
classifying, by the defect detector module, one or more first features from the image of the physical object into one or more first classifications using one or more machine learning algorithms;
analyzing, by the defect detector module, the one or more first classifications;
determining, by the defect detector module, that the physical object comprises a defect based on analyzing the one or more first classifications
classifying, by the defect detector module, one or more second features from the image of the physical object into one or more second classifications using the one or more machine learning algorithms, wherein the one or more second classifications comprise at least one of the following: a bar, a discontinuity, and an end post; and
cropping, by the defect detector module, the image to an area surrounding the one or more second features to create a cropped image;
wherein classifying the one or more first features from the image of the physical object into the one or more first classifications comprises classifying the one or more first features from the cropped image of the physical object into the one or more first classifications.
16. The one or more non-transitory computer-readable storage media of claim 15 , the operations further comprising:
determining, by the defect detector module, a location of the defect within the image;
determining, by the defect detector module, that the location of the defect within the image is part of the physical object; and
determining, by the defect detector module, a geographic location of the defect of the physical object based at least in part on the location of the defect within the image.
17. The one or more non-transitory computer-readable storage media of claim 15 , the operations further comprising labeling, by the defect detector module, the one or more first features of the image with one or more labels.
18. The one or more non-transitory computer-readable storage media of claim 15 , the operations further comprising training the one or more machine learning algorithms to classify the one or more first features from the image by collecting sample data representative of the one or more first features.
19. The one or more non-transitory computer-readable storage media of claim 15 , wherein:
the physical object is a rail joint;
the defect is a broken rail joint; and
the one or more first classifications comprise one or more of the following:
a bolt;
a break;
a hole; and
a discontinuity.
20. The one or more non-transitory computer-readable storage media of claim 15 ,
wherein the one or more second classifications comprise at least one of the following: a bar, a discontinuity, and an end post.