Research on DWTT Fracture Evaluation Based on U-net Network Model

In view of the characteristics of the difficulty and low efficiency of the fracture assessment method of the drop hammer tear test at home and abroad, this paper proposes a semantic segmentation method for steel fractures based on up-down sampling and cross-layer connection. First, the data of the collected sample fractures is made into a data set. Under CentOS operation system, Python3.6 and TensorFlow-1.12.0 deep learning frameworks are built, and the U-net network model is used for training. Use the training model to predict the data set, and manually evaluate the prediction results. After manual evaluation of the experimental results, it is shown that the method proposed in this paper can accurately segment the DWTT fracture image.