Team TRM
Team Contributors: Trung, Rosemond, Michael, & Nikita
Python 3.6+ is required for all versions of this project.
The final version of the project is run using the fastai library. Install instructions can be found on their github page: https://github.com/fastai/fastai
Previous versions require the following libraries: numpy pytorch sklearn PIL tqdm pandas
Most of these libraries are installed with anaconda automatically, which is the suggested method of install for all of the packages as it handles dependecies gracefully.
Initially, our python files MUST be run in the order of train.py and then test.py. Train.py will generate a model file and save it to the working directory after it is run. After you are happy with the trained network only the test.py will be utilized. Train.py and test.py MUST be in the same directory.
Before running train.py: Create this directory structure in the same path as the train.py and test.py:
data
train
clas1
clas2
...
valid
clas1
clas2
...
test
clas1
clas2
...
models
testAF
clas1
clas2
...
Place the easy test data in the testAF directory in the hard test data in the test directory
Warning:
Please close your file explorer after placing the files in the necessary directories to avoid error.
Running train.py: If the above path is adhered to, no additional steps are needed to run train.py.
Running test.py: If the above path is adhered to, no additional steps are needed to run test.py.
Results: The predictions will be in an estimatedEasyLabels.txt file and estimatedHardLabels.txt file accordingly.