Application of deep learning for evaluation of the growth rate of Daphnia magna.

Abstract

For the safe use of chemicals widely used in human activities, it is crucial to assess their ecological impacts when released into the environment. Daphnia, a well-established environmental indicator species, is commonly used to evaluate the biological effects of chemicals and testing methods have been established. Among various indicators, the growth rate is one of the important parameters, but it requires significant time and effort to measure. In this study, we applied deep learning-based image recognition techniques to extract images of Daphnia from live imaging and assess their size. The estimated size of Daphnia, derived from images processed through deep learning, showed a high correlation with measured values, demonstrating the capability to measure Daphnia size from the images while they are swimming. This approach enables non-invasive measurements of Daphnia size without complicated procedures, which not only streamlines ecological impact assessments but also presents a valuable technique for ecological studies, such as analyzing the size distribution of zooplankton.