During the services of engineering materials and structural components, all kinds of corrosion, fatigue, fracture, wear, and other failures will occur due to not only the inappropriate selection of materials and structures but also the corrosive medium and loading of the external environment. One of the key basic technologies of failure analysis is failure image diagnosis. The current failure image intelligent diagnosis technology has become a research hotspot, and the development and application of high-precision image segmentation technology is a key path to enhance failure mechanism identification and failure analysis. In this paper, the research progress of image segmentation technologies is mainly discussed. Both the working principles, technical characteristics, and deficiencies of traditional image segmentation technology and deep learning-based image segmentation technology are outlined. The important roles of deep learning models such as CNN, FCN, U-Net, etc. in improving the precision and accuracy of image segmentation are elaborated in detail. In the end, the prospects for the application of image segmentation technology in the field of engineering failure analysis are highlighted.