クラス分類のための細線化処理とダイクストラのアルゴリズムを用いた物体の中心推定法

抄録

In the medical field, several studies have been conducted to elucidate the characteristics of human cells using microorganisms that have the same functions as human cells. To utilize microorganisms, it is necessary to cultivate microorganisms efficiently. The existence of eggs is an important indicator for assessing the growth of microorganisms. The purpose of this research is to automatically determine the presence or absence of microbial eggs. In this study, since it is necessary to reduce the number of pixels from high-resolution images, the presence or absence of eggs is classified by cutting out the area around the body center of the microorganism and inputting it into EfficientNet. To estimate the center of the body of the microorganism, we consider applying Dijkstra's algorithm to a single pixel wide image obtained by the thinning process. The shortest path between any two end points of the narrowing shape is calculated using Dijkstra's algorithm. The midpoint of the longest shortest path is taken as the body center.