| abstract |
A method for identifying and extracting free liquid surface based on a U-net convolutional neural network model belongs to the technical field of image processing and automatic detection. First, the liquid level image is processed and the free liquid level is manually marked to form a dataset with segmented liquid level images. The dataset is divided into three parts: training set, validation set and test set. Second, build the U-net convolutional neural network model; import the training set into the U-net convolutional neural network, perform feature learning on the image, and predict the free surface. Again, use the validation set to validate the model and tune the model, saving the optimal model when the loss function is not decreasing. Finally, the trained U‑net convolutional neural network model is deployed for automatic detection of liquid level images in the test set and evaluation of the model. The invention can extract the free liquid surface of the breaking wave through the method of combining non-contact measurement and artificial intelligence, and can solve the technical problem that the liquid surface identification of the breaking wave is difficult. |