Abstract: An information processing device includes a first determination section and a first output section. First output data is obtained from a trained model, which has being trained beforehand, by input of first input data to the trained model. Second output data is obtained from the trained model by input of second input data to the trained model, in which second input data a perturbation of a specified perturbation amount is applied to the first input data. The first determination section makes a determination as to whether a change amount of the second output data relative to the first output data is not more than a pre-specified threshold. The first output section outputs information representing the first input data and a perturbation amount for which the change amount is determined by the first determination section to be not more than the threshold.

Abstract: An information processing device includes a control unit that performs prediction, from a first movement route including a predetermined number of continuous road links, of one or more road links that are traveled by a first mobile object subsequent to the first movement route, based on movement history information on one or more mobile objects traveling on a road. The control unit performs the prediction by using a machine learning model that, for an input of sequential data, outputs one piece of data or sequential data subsequent to the input sequential data.

Abstract: A pattern updating device that includes a memory, and a processor coupled to the memory. The processor is configured to extract frequently-occurring patterns from a plurality of path information, store the extracted patterns as a predetermined data structure, perform a check of newly acquired path information against the stored predetermined data structure, and based on the check, update the frequently-occurring patterns in all path information, including the newly acquired path information, by updating the stored predetermined data structure.

Abstract: The retrieval device extracts a feature corresponding to search text by inputting the search text into a pre-trained text feature extraction model. The retrieval device then, for plural combinations stored in a database associating a text description including plural sentences, with a vehicle-view video, and with vehicle behavior data representing temporal vehicle behavior, computes a text distance represented by a difference between a feature extracted from each sentence of the text description associated with the video and vehicle behavior data, and the feature corresponding to the search text. The retrieval device outputs as the search result a prescribed number of pairs of video and vehicle behavior data pairs in sequence from the smallest text distance according to the text distances.

Abstract: A danger prediction device includes a processor. The processor is configured to acquire, from a traveling vehicle, position information for the traveling vehicle on a travel path and behavior information for the traveling vehicle at a location corresponding to the position information, compile, from a plurality of acquired items of behavior information, behavior information corresponding to position information for locations having a similar attribute, and input compiled behavior information to a prediction model generated based on pre-gathered vehicle behavior information and a danger level corresponding to the pre-gathered vehicle behavior information, and perform danger prediction for a location corresponding to the compiled behavior information.

Abstract: An information processing device includes a first determination section and a first output section. First output data is obtained from a trained model, which has being trained beforehand, by input of first input data to the trained model. Second output data is obtained from the trained model by input of second input data to the trained model, in which second input data a perturbation of a specified perturbation amount is applied to the first input data. The first determination section makes a determination as to whether a change amount of the second output data relative to the first output data is not more than a pre-specified threshold. The first output section outputs information representing the first input data and a perturbation amount for which the change amount is determined by the first determination section to be not more than the threshold.

Abstract: The retrieval device extracts a feature corresponding to search text by inputting the search text into a pre-trained text feature extraction model. The retrieval device then, for plural combinations stored in a database associating a text description including plural sentences, with a vehicle-view video, and with vehicle behavior data representing temporal vehicle behavior, computes a text distance represented by a difference between a feature extracted from each sentence of the text description associated with the video and vehicle behavior data, and the feature corresponding to the search text. The retrieval device outputs as the search result a prescribed number of pairs of video and vehicle behavior data pairs in sequence from the smallest text distance according to the text distances.

Abstract: An image processing device includes: an extraction unit that performs a convolution processing and a pooling processing on information of an input image including an image of a person and extracts a feature from the input image to generate a plurality of feature maps; a first fully connected layer that outputs first fully connected information generated by connecting the plurality of feature maps; a second fully connected layer that connects the first fully connected information and outputs human body feature information indicating a predetermined feature of the person; and a third fully connected layer that connects the first fully connected information or the human body feature information to output behavior recognition information indicating a probability distribution of a plurality of predetermined behavior recognition labels.