The mapping of faults provides essential information on many aspects of seismic exploration, characterisation of reservoirs for compartmentalisation and cap-rock integrity. However, manual interpretation of faults from seismic data is time-consuming and challenging due to limited resolution and seismic noise. In this study, we apply a convolutional neural network trained on synthetic seismic data with planar fault shapes to improve fault mapping in the Lower and Upper Cretaceous sections of the Valdemar Field in the Danish North Sea. Our objective is to evaluate the performance of the neural network model on post-stack seismic data from the Valdemar Field. Comparison with variance and ant-tracking attributes and a manual fault interpretation shows that the neural network predicts faults with more details that may improve the overall geological and tectonic understanding of the study area and add information on potential compartmentalisation that was previously overlooked. However, the neural network is sensitive to seismic noise, which can distort the fault predictions. Therefore, the proposed model should be treated as an additional fault interpretation tool. Nonetheless, the method represents a state-of-the-art fault mapping tool that can be useful for hydrocarbon exploration and CO2 storage site evaluations.