The identification and monitoring of coastline and shoreline plays an important role in coastal erosion assessment. The deep learning models can be a potential tool to detect the coastlines and shorelines in Vietnam using ultra-high resolution satellite images. The aims of the study are: i) To propose a set of indicators to determine the coastlines and shoreline; ii) To build deep machine learning models that automatically interpret the coastlines and shorelines on ultra-high resolution remote sensing images; and iii) To apply developed deep learning (DL) models to monitor coastal erosion in central Vietnam. Eight DL models were implemented based on four artificial intelligence network structures, including U-Net, U2-Net, U-Net3+, and DexiNed. Satellite images collected through Google Earth Pro software were used as input for all models. As a result, the U-Net model has been effectively applied to coasts in Cu De, Lai Giang, and Bien Lo estuaries,. The output results were used to calculate the rate of erosion/accretion in these areas. Additionally, the study indicated that coastline is a suitable criterion in assessing coastal erosion under the impact of sea level rise during storms. On the other hand, shoreline is a suitable criterion in assessing tidal fluctuations or instantaneous movements of wave currents during the year.