Abstract:

In order to solve the problem of slime water overflow in the factory, a solution of real-time monitoring of slime water overflow was put forward. First, CNN was used to extract features from the slime water monitoring images, and then the extracted features were sent into the model for training. The accuracy rate was continuously improved through fine tuning. Finally, relevant evaluation indicators were used to evaluate the performance of the model. The results show that compared with the traditional detection algorithm, the slime water overflow detection model based on the CNN has an improvement of 15% in terms of evaluation indicators. Accurate judgment can also be made on the severity of slime water overflow, and all evaluation indicators exceed 90%, which is helpful to reduce the occurrence of slime water overflow.

Key words: convolutional neural network (CNN), slime water overflow, video detection, feature extraction, evaluation indicator