Abstract: An Objective of the present invention is to improve the inter-camera cooperation in a monitoring system comprising multiple cameras each having a function to perform video analysis by use of an AI model. A monitoring system according to the present example comprises: multiple AI cameras 101 each having a function to perform video analysis by use of an AI mode; and an analyzing device 102 that generates the AI model used in the AI cameras 101. Each of the AI cameras 101 detects suspicious behaviors through video analysis using a particular-behavior detection AI model for detecting the suspicious behaviors. When suspicious behaviors are detected by any of the AI cameras 101, the analyzing device 102 generates a particular-person detection AI model for detecting a particular person (suspicious person) having done the suspicious behaviors, and transmits the particular-person detection AI model to each of the AI cameras 101.

Abstract: Even if an existing learning dataset is limited, a new learning dataset with sufficient variation is generated. Therefore, for each of a plurality of learning data subsets, new input signals are generated from input signals of a plurality of pieces of learning data, and a plurality of pieces of new learning data that are respectively combinations of the new input signals and output signals of the corresponding learning data subset are generated. The input signals of the plurality of pieces of the learning data included in the corresponding learning data subset are divided into a first signal group and a second signal group, and the new input signals are generated by a learning device that is generated by performing learning by the first signal group set as an input signal set and the second signal group set as an output signal set.

Abstract: The present invention provides a technique suitable for the detection of small objects such as white canes. In order to solve this problem, one representative example of this image processing system comprises a video acquisition unit configured to input an input image acquired as a video, an estimation unit including a Fully Convolutional Network (FCN) learning model trained with respect to a region of a predetermined detection target, and configured to generate a region estimation image for the detection target by processing the input image with the learning model, and an image processing unit configured to detect the region of the detection target based on the region estimation image.

Abstract: The present invention provides a technique suitable for the detection of small objects such as white canes. In order to solve this problem, one representative example of this image processing system comprises a video acquisition unit configured to input an input image acquired as a video, an estimation unit including a Fully Convolutional Network (FCN) learning model trained with respect to a region of a predetermined detection target, and configured to generate a region estimation image for the detection target by processing the input image with the learning model, and an image processing unit configured to detect the region of the detection target based on the region estimation image.

Abstract: The cloud server holds the main database for storing all the registration data that is handled in the present system, and the edge server that is arranged close to the sensor holds a sub-database for storing part of the registration data. The sub-database in the edge server stores only the registration data having a high probability of being verified in the edge server. In the case where the edge server verifies the detection data that has been acquired by the sensor with the registration data within the sub-database and determines that the registration data that matches the detection data does not exist within the sub-database, the configuration allows the detection data to be transmitted to the cloud server and requests the detection data to be verified with the registration data within the main database.

Abstract: Even if an existing learning dataset is limited, a new learning dataset with sufficient variation is generated. Therefore, for each of a plurality of learning data subsets, new input signals are generated from input signals of a plurality of pieces of learning data, and a plurality of pieces of new learning data that are respectively combinations of the new input signals and output signals of the corresponding learning data subset are generated. The input signals of the plurality of pieces of the learning data included in the corresponding learning data subset are divided into a first signal group and a second signal group, and the new input signals are generated by a learning device that is generated by performing learning by the first signal group set as an input signal set and the second signal group set as an output signal set.