Author: Zhaohui Che E-mail: [email protected]
This code is for Blind Quality Estimation and Distortion Classification Task of Industrial 2D Matrix Symbols
There are 4 .py files in total
How to RUN the code?
- You should download the training and testing database given in the paper first.
- You should change the data path as your own path, please follow the hints of the .py files
- You should Run "Train_Subtask_1.py" to train the 1st sub-network for distortion type Classification Task
- You should Run "Train_Subtask_2.py" to inherit the weights and biases of the shared layers obtained by step.3, and further train the 2nd sub-network for quality grade eitimation task.
- You should Run "Test_Subtask_1.py" to test the distortion classification accuracy of the proposed model on Test Database.
- You should Run "Test_Subtask_2.py" to predict the quality grade on Test Database, and the SRCC, PLCC, RMSE metrics are provided in "Metrics" folder for comparing performance.
If you have any question, please contact me.
Zhaohui Che 2018.1.31