Contributors: Michael Butler, Preston Ching, M. Elaine Cunha
To run the full model, type python run.py into your command line. This assumes you're running the pipeline in an environment with tensorflow v1. Detailed Replication instructions to run on AWS in Replication.md.
Directory
- Train/: folder with subdirectories for labeled and unlabeled training data
- Figures/: folder containing results from performance tests
- Old/: folder containing out-of-date versions of files
- architecture.py: defines CNN architecture
- config.py: configures data parameters (i.e., image resize dimensions)
- run.py: trains one instance of the supervised or semisupervised model
- utils.py: contains functions for image analysis and developing training/testing sets
- semisupervised.py: specifies training methodology for the semisupervised model
- supervised.py: specifies training methodology for the fully supervised model