Tracking method of medaka considering proximity state

In biology, it is expected to understand the mechanism of herd development by analyzing crowd behavior. Currently, medaka (Oryzias latipes) is often used when conducting such research because of its ease of breeding and experimentation. However, when conducting such studies, it is very difficult to manually obtain herd data. In this article, in order to analyze the behavior of medaka, we will introduce a method to automatically extract the swimming trajectory of medaka from time-series images and acquire the data necessary for behavior analysis. In particular, the improvement of extraction accuracy is achieved by processing the overlap state and adjoin state of medaka in the image.