Learning data generation method / program for event occurrence estimation, learning model and event occurrence estimation device

摘要

PROBLEM TO BE SOLVED: To provide a method of generating learning data for generating a learning model capable of estimating the susceptibility to a predetermined event based on input image data relating to a situation of a surrounding or a predetermined environment. SOLUTION: In this learning data generation method, a process of assigning a value related to the situation to each pixel or a unit area is performed on image data related to the situation of the surrounding or a predetermined environment, and the situation is assigned to each formed pixel area. Based on the step of generating the area-divided image data which is the image data with the label according to the above, and the pixel area and the label of the area-divided image data, the situation information which is the information which can be related to the occurrence of a predetermined event is determined. It has a step and a step of generating training data including a data set in which a value indicating that a predetermined event is likely to occur is associated with a value indicating that a predetermined event is likely to occur with respect to image data for which status information satisfying a predetermined condition is determined as correct answer data. doing. [Selection diagram] Fig. 1