christinaliang/fp_bme6938

There are two folders inside the whole profile, and data is saved in data_bcidr folder, with the traing and testing images. the image set is partial, and I only leave 170 samle images in the training folder. As models folder and data folder are big, so I upload them into dropbox and can be downloaded by the below url: https://www.dropbox.com/sh/fv2tsxybsa06in8/AACr1IYJYrC92Ss9kURELiysa?dl=0

1. Preprocessing

This step is used for installing the required packages.

$ pip install -r requirements.txt   

2. Building vocab

This step has been done for generating vocab , and the vocab.pkl is saved inside the data folder.

$ python build_vocab.py   

3. Train the model

Can just run train.py for training setp. The default path is saved in args function, and if needed they can be changed.

$ python train.py    

4. Test the model

It will run the testing process for the images in the testing foler. The result will be saved as one csv file in the root folder.

$ python test.py