Structure defect detection using machine learning algorithms

A computer-implemented method of analyzing an image of a surface to detect a defect in the surface, comprising:

• receiving the image of the surface;
• processing the image using a machine learning algorithm configured to detect the defect, the machine learning algorithm comprising a convolutional neural network including:
• at least one convolution layer;
• and
• at least one max pooling layer;
• and displaying the image with location of the defect being indicated if determined to be present by the convolutional neural network;
• wherein the convolutional neural network comprises:
• an input layer having a height of n pixels, a width of n pixels, and a depth of d channels;
• said at least one convolution layer comprising a preliminary convolution layer, a secondary convolution layer, a tertiary convolution layer and a quaternary convolution layer;
• said at least one pooling layer comprising a preliminary pooling layer and a secondary pooling layer;
• the preliminary convolution layer having a height of Hc1 pixels, a width of Hc1 pixels, and a depth of Dc1 channels formed by a first convolution operator having a height of hc1 pixels, a width of hc1 pixels, and a depth of dc1 channels with a stride of sc1 performed upon the input layer;
• wherein Hc1=[(n−hc1)/sc1]+1;
• wherein Dc1=dc1;
• the preliminary pooling layer having a height of Hp1 pixels, a width of Hp1 pixels, and a depth of Dp1 channels formed by a first pooling operator having a height of hp1 pixels and a width of hp1 pixels with a stride of sp1 performed on the preliminary convolution layer;
• wherein Hp1=[(Hc1−hp1)/sp1]+1;
• wherein Dp1=Dc1;
• the secondary convolution layer having a height of Hc2 pixels, a width of Hc2 pixels, and a depth of Dc2 channels formed by a second convolution operator having a height of hc2 pixels, a width of hc2 pixels, and a depth of dc2 channels with a stride of sc2 performed upon the preliminary pooling layer;
• wherein Hc2=[(Hp2−hc2)/sc2]+1;
• wherein Dc2=dc2;
• the secondary pooling layer having a height of Hp2 pixels, a width of Hp2 pixels, and a depth of Dp2 channels formed by a second pooling operator having a height of hp2 pixels and a width of hp2 pixels with a stride of sp2 performed upon the secondary convolution layer;
• wherein Hp2=[(Hc2−hp2)/sp2]+1;
• wherein Dp2=Dc2;
• the tertiary convolution layer having a height of Hc3 pixels, a width of Hc3 pixels, and a depth of Dc3 channels formed by a third convolution operator having a height of hc3 pixels, a width of hc3 pixels, and a depth of dc3 channels with a stride of sc3 that is performed upon the secondary pooling layer;
• wherein Hc3=[(Hp2−hc3)/sc3]+1;
• wherein Dc3=dc3;
• an activation layer having a height of Ha1 pixels, a width of Ha1 pixel, and a depth of Da1 channels formed by a nonlinear activation function operator performed upon the tertiary convolution layer;
• wherein Ha1=Hc3;
• wherein Da1=Dc3;
• the quaternary convolution layer having a height of Hc4 pixels, a width of Hc4 pixels, and a depth of Dc4 channels formed by a fourth convolution operator having a height of hc4 pixel, a width of hc4 pixel, and a depth of dc4 channels with a stride of sc4 performed upon the activation layer;
• wherein Hc4=[(Ha1−hc4)/sc4]+1;
• wherein Dc4=dc4;
• and a softmax layer having a height of Sm1 pixels, a width of Sm1 pixels, and a depth of Dsm1 channels formed by a softmax operator performed upon the quaternary convolution layer such that a continuously extending line in an image can be detected;
• wherein Sm1=Hc4;
• wherein Dsm1=Dc4.