Coating region learning device, coating region learning method, coating region prediction device, and coating region prediction method

An application area learning method in which a computer is configured to perform machine learning on a correlation between a feature value at least inside a contour extracted from an outline image of a bottom surface of an upper portion of a shoe and an application area of an adhesive in the upper portion of the shoe and generates a prediction model;

• the machine learning including a multilayer neural network and weights of connections between layers of the neural network, the weights learned by deep learning so that an error between application area data of correct answer data and output data output from an output layer of the neural network is minimized;
• the minimizing is performed by adjusting the weights of each layer of the neural network so that the error supplied from the output layer is minimized;
• the learned neural network by which the error with respect to the feature value within the contour and the application area data has been minimized is stored for estimating the application area from the feature value within the contour extracted from the outline image of an unknown shoe component, in an application area prediction device.

An application area prediction method comprising, based on a prediction model generated by performing machine learning on a correlation between a feature value at least inside a contour extracted from an outline image of a bottom surface of an upper portion of a shoe and an application area of an adhesive in the upper portion of the shoe, predicting with a computer the application area from the feature value at least inside the contour extracted from the outline image of a shoe component and outputs outputting the application area;

• the machine learning including a multilayer neural network and weights of connections between layers of the neural network, the weights learned by deep learning so that an error between application area data of correct answer data and output data output from an output layer of the neural network is minimized;
• the minimizing is performed by adjusting the weights of each layer of the neural network so that the error supplied from the output layer is minimized;
• the learned neural network by which the error with respect to the feature value within the contour and the application area data has been minimized is stored for estimating the application area from the feature value within the contour extracted from the outline image of an unknown shoe component, in an application area prediction device.

A non-transitory computer readable medium storing an application area learning program causing a computer to perform machine learning on a correlation between a feature value at least inside a contour extracted from an outline image of a bottom surface of an upper portion of a shoe and an application area of an adhesive in the upper portion of the shoe and generate a prediction model;

• the machine learning including a multilayer neural network and weights of connections between layers of the neural network, the weights learned by deep learning so that an error between application area data of correct answer data and output data output from an output layer of the neural network is minimized;
• the minimizing is performed by adjusting the weights of each layer of the neural network so that the error supplied from the output layer is minimized;
• the learned neural network by which the error with respect to the feature value within the contour and the application area data has been minimized is stored for estimating the application area from the feature value within the contour extracted from the outline image of an unknown shoe component, in an application area prediction device.

A non-transitory computer readable medium storing an application area prediction program causing a computer to predict, based on a prediction model generated by performing machine learning on a correlation between a feature value at least inside a contour extracted from an outline image of a bottom surface of an upper portion of a shoe and an application area of an adhesive in the upper portion of the shoe, the application area from the feature value at least inside the contour extracted from the outline image of a shoe component and to output the application area;

• the machine learning including a multilayer neural network and weights of connections between layers of the neural network, the weights learned by deep learning so that an error between application area data of correct answer data and output data output from an output layer of the neural network is minimized;
• the minimizing is performed by adjusting the weights of each layer of the neural network so that the error supplied from the output layer is minimized;
• the learned neural network by which the error with respect to the feature value within the contour and the application area data has been minimized is stored for estimating the application area from the feature value within the contour extracted from the outline image of an unknown shoe component, in an application area prediction device.