The spatiotemporal variability of the major phytoplankton groups, such as dinoflagellates and diatoms, provides crucial information about the ecosystem’s status, especially when it comes to coastal regions that are influenced by permanent anthropogenic pressure. The purpose of this study was to develop specific models for the retrieval of diatoms and dinoflagellates in Annaba Bay and El Kala’s coast (the eastern part of the Algerian coast). We established a data set that included quantified micro-phytoplankton densities from seawater samples obtained at distinct stations along the study area during different seasons and their corresponding Sentinel-3 and MODIS reflectance (Rrs) values. Several band ratios based on the blue-green and near infrared-red (NIR-red) parts of the spectrum have been tested, as well as the Generalized Linear Model (GLM) with 12 bands using the machine learning approach, in order to validate the most accurate model for quantifying micro-phytoplankton density. The results revealed the efficiency of the 6B.S3 (r = 0.90, RMSE = 3364.2 Cells l-1) based on the band ratio, which employs six bands in the blue-green and red parts of the spectrum, and the 12B.S3 (r = 0.84, RMSE = 410.49 Cells l-1) based on the machine learning approach, which employs 12 bands of the spectrum for the estimation of diatom and dinoflagellate densities, respectively. On the other hand, both groups exhibit strong correlation with the 5B.M, which involves 5 bands of MODIS Rrs (r = 0.90 and 0.76 for diatoms and dinoflagellates, respectively). Mapping phytoplankton densities revealed that Sentinel-3 data and models outperformed those of MODIS and were more suitable for monitoring diatoms and dinoflagellates along Algeria’s eastern coast.