The article introduces a new dataset called DAPlankton that is aimed at developing and benchmarking domain adaptation methods for image recognition. Different imaging instruments and local differences in plankton species composition have caused a large pool of varying plankton image datasets. DAPlankton consists of phytoplankton images captured using different imaging instruments. It includes two subsets: DAPlankton LAB with images of cultured phytoplankton, and DAPlankton SEA with images collected from the Baltic Sea. The paper also provides a benchmark comparison of three widely used domain adaptation methods.
Publication date: 8 Feb 2024
Project Page: https://arxiv.org/abs/2402.05615v1
Paper: https://arxiv.org/pdf/2402.05615
