Pixel-wise phase unwrapping of fringe projection profilometry based on deep learning

•This paper presents a deep learning-based phase unwrapping method that achieves pixel-wise phase unwrapping using only N + 2 fringe patterns.•The paper presents a method to create data sets for network training and verification.•The effectiveness and accuracy of the method have been validated in scenarios involving complex topological structures, fringe patterns with different exposures, and varying noise levels.