Abstract: Preparing actual defective product data and non-defective product data, causing a deep generation model to learn the data and to generate latent variables in which features of the non-defective product data and the actual defective product data are mixed, causing a classification model to learn the latent variables to generate a classified non-defective product and defective product latent variable, deleting the non-defective product latent variable from the classified non-defective product and defective product latent variable to output the defective product latent variable including a gray latent variable, causing a distance learning model to learn the defective product latent variable and the non-defective product latent variable and to delete the gray latent variable, and causing the deep generation model to learn the defective product latent variable that has been obtained and to generate the pseudo defective product data greater in number than the actual defective product data.

Abstract: A teacher data preparation method for preparing non-defective product teacher data and defective product teacher data, in a defect classification model that performs learning using a non-defective product image and a defective product image, the teacher data preparation method comprising: acquiring, as the non-defective product teacher data, many pieces of non-defective product feature quantity data obtained by extracting a feature quantity in a predetermined first number of dimensions from a large number of the non-defective product images; and acquiring, as the defective product teacher data, many pieces of generated defective product feature quantity data obtained by generating the feature quantity in the predetermined first number of dimensions, by using a generation model that has performed learning using defective product feature quantity data obtained by extracting the feature quantity in the predetermined first number of dimensions from the defective product images smaller in number than the non-de

Abstract: Preparing actual defective product data and non-defective product data, causing a deep generation model to learn the data and to generate latent variables in which features of the non-defective product data and the actual defective product data are mixed, causing a classification model to learn the latent variables to generate a classified non-defective product and defective product latent variable, deleting the non-defective product latent variable from the classified non-defective product and defective product latent variable to output the defective product latent variable including a gray latent variable, causing a distance learning model to learn the defective product latent variable and the non-defective product latent variable and to delete the gray latent variable, and causing the deep generation model to learn the defective product latent variable that has been obtained and to generate the pseudo defective product data greater in number than the actual defective product data.

Abstract: Included are a first feature quantity conversion unit that converts a plurality of pieces of acquired actual defective product data respectively into actual feature quantities, a predicted feature quantity generation unit that generates a predicted feature quantity group by learning of a feature quantity generation model, a pseudo defective product data generation unit that generates a pseudo defective product data group by learning of an image generation model, a second feature quantity conversion unit that converts the pseudo defective product data group to acquire as a pseudo feature quantity group, a feature quantity distribution comparison unit that compares distributions between the predicted feature quantity group and the pseudo feature quantity group to calculate a feature quantity error as a residual error, and a pseudo defective product data quality determination unit that determines the quality of the generated pseudo defective product data group, based on the feature quantity error.

Abstract: An autoencoder is caused to learn based on data of a majority class of imbalanced data. The imbalanced data is input to the autoencoder that has learned, and an error in a predetermined parameter between an input image data and an output image data is acquired. Data of each class is extracted as teacher data such that the number of pieces of first class data counted in descending order of the size of the obtained error among the first class data and the number of pieces of second class data are balanced. A convolutional neural network is caused to learn the teacher data to obtain a learning model.

Abstract: A teacher data generating method includes: causing a generation model to perform learning by using a learning image in which boundary information indicating in a chromatic color a range of a defect is superimposed on a defective product image in a gray scale so as to generate a generated defect image including a new image of a defect in the gray scale and an image of the boundary information in the chromatic color; generating the generated defect image by using the generation model; extracting a pixel having a pixel value corresponding to the chromatic color from the generated defect image, extracting the boundary information corresponding to the generated defect image, and acquiring a gray scale defect image without including an image of the boundary information; calculating a coordinate of the boundary information; and associating the gray scale defect image with the coordinate to obtain defective product teacher data.

Abstract: A first standard deviation ? of feature quantities of all pieces of non-expert data stored in a non-expert data storage unit 13 is calculated, and a second standard deviation ? of feature quantities of all pieces of expert data stored in an expert data storage unit 14 is calculated. In addition, a first rank sum ? of the feature quantities of all pieces of non-expert data stored in the non-expert data storage unit 13 is calculated, and a second rank sum ? of the feature quantities of all pieces of expert data stored in the expert data storage unit 14 is calculated. Then, a continuation and an end of acquisition of defective product data by an expert are determined, based on the first standard deviation ? and the second standard deviation ?, and the first rank sum ? and the second rank sum ?.

Abstract: An inspection device 1 that inspects whether an image to be inspected is a normal image or an abnormal image includes a learning unit 20A that learns so as to enable reconstruction of normal image data, an error calculation unit 23 that calculates a reconstruction error when a plurality of pieces of input image data are input to the learning unit 20A, a threshold calculation unit 24 that calculates a threshold based on the reconstruction error, and an identification unit 25 that performs identification based on the threshold. The error calculation unit 23 sequentially compares input image data DI and output image data DO while scanning a detection area E on an image data.

Abstract: The present invention relates to a collecting method for training data for collecting defective product data as training data for learning a classification model that classifies the inspected object as a normal product or an abnormal product, including: collecting many pieces of defective product data (step 1 of FIG. 3); extracting a plurality of feature quantities respectively from many pieces of the defective product data (step 2); calculating a state sum for every feature quantity of the plurality of feature quantities that have been extracted, for the many pieces of the defective product data (step 3); calculating, as an index value, a logarithmic sum of a plurality of state sums that have been calculated (step 4); and ending collecting the defective product data, in a case where the calculated index value is equal to or greater than a predetermined target value (step 5).

Abstract: A feature quantity selection method according to the present invention includes extracting a multidimensional feature quantity from expert data including images of various defect shapes, extracting a multidimensional feature quantity from non-expert data including images of limited defect shapes, calculating a standard deviation in every dimension of the respective feature quantities of the expert data and the non-expert data that have been extracted; calculating a ratio between the standard deviation in every dimension of the calculated expert data and the standard deviation in every dimension of the calculated non-expert data, selecting a predetermined number of standard deviation ratios in descending order of value from among the standard deviation ratios that have been calculated, and selecting feature quantities associated with the standard deviation ratios that have been selected, as the feature quantities each having a high contribution degree to a specific defect mode.

Abstract: A teacher data collecting method in a defect classification model for classifying a defect by using, as teacher data, a few pieces of expert data and many pieces of non-expert data, includes: encoding, into one dimension, a latent variable of a variational auto encoder that has been caused to perform learning the expert data; inputting the non-expert data into the variational auto encoder and encoding a latent variable into one dimension; calculating maximum values and minimum values of the latent variable in one dimension of the expert data and the non-expert data; and determining whether to complete collection of the non-expert data, based on a ratio of a difference between the maximum value and the minimum value of the latent variable in one dimension of the non-expert data to a difference between the maximum value and the minimum value of the latent variable in one dimension of the expert data.

Abstract: A pinch detection device at an opening/closing section has a very simple structure that can surely detect a pinch without faulty operation in a vehicle whose power supply voltage greatly fluctuates. The pinch detection device includes a sensor for detecting displacement of an opening/closing section of a vehicle to be opened/closed by a motor, and a time counter for counting a relative time adjusted according to a fluctuation in a power supply voltage. The pinch detection device also has a pinch detector for detecting a pinch of a foreign matter at the opening/closing section according to a change in a relative speed obtained based on a change in the displacement on a basis of the relative time.

Abstract: A control device at an opening/closing section of a vehicle and a method for controlling the opening/closing section of the vehicle control a motor so that a pinch at the opening/closing section is determined accurately. The control device includes a pinch determination device that is mounted to the motor for opening/closing the opening/closing section of the vehicle and determines a pinch of a foreign object based on a change in the rotation number of the motor, and demagnetization pulse applying device for supplying power with reverse polarity to the motor in a pulse-like manner at the completion of the opening/closing.

Abstract: An automatic control system synchronizes and controls a control target while preventing unnecessary signal processing. The system includes a slave station that is connected to the control target, a master station that is connected to a control unit for controlling the control target, a contact point information collection and distribution device that collects and distributes a series of contact point information communicated at a predetermined communication cycle, and communication lines connecting between the master and slave stations via the contact point information collection and distribution device. The contact point information collection and distribution device collects and distributes the series of contact point information in synchronization with all the communication lines under the control of the control unit.

Abstract: The present invention provides a control device at an opening/closing section of a vehicle and a method for controlling the opening/closing section of the vehicle for controlling a motor so that pinch at the opening/closing section is determined accurately while its constitution is very simple. The control device includes a pinch determination device that is mounted to a motor for opening/closing an opening/closing section of a vehicle and determines pinch of a foreign matter based on a change in the rotation number of the motor, and demagnetization pulse applying means for supplying power with reverse polarity to motor in a pulse-like manner at completion of the opening/closing.

Abstract: The present invention provides a pinch detection device at an opening/closing section having a very simple structure that can surely detect pinch without a faulty operation also in a vehicle whose power supply voltage greatly fluctuates, a vehicle having this device, and a method for detecting pinch at an opening/closing section. This invention is a pinch detection device at an opening/closing section includes a sensor for detecting displacement of a opening/closing section of a vehicle to be opened/closed by a motor, a time counter for counting a relative time adjusted according to a fluctuation in a power supply voltage, and a pinch detector for detecting pinch of a foreign matter at the opening/closing section according to a change in a relative speed obtained based on a change in the displacement on a basis of the relative time.