Underwater target detection learning device and underwater target detection device

摘要

PROBLEM TO BE SOLVED: To provide a system for detecting an underwater target using a neural network, operating a large number of learning models depending on observation conditions, and realizing highly accurate target detection. SOLUTION: A learning model aggregate 114 in which a plurality of learning models related to observation data of an underwater target are accumulated together with the observation condition 101 is created, a distance calculation means 109 for calculating the distance between the observation conditions 101 is provided, and the learning model aggregate is provided. Based on the distance between the conditions of 114, the model selection means 108 selects one or more corresponding learning models, and the inference means 122 estimates the underwater target by the extracted learning model. [Selection diagram] Fig. 1