Running instructions

To run in the default settings please follow the following steps:

1. Install the dependencies listed in ./requirements.txt
2. Paste the idd20k_lite.zip file in the ./data ditectory.
3. Go into the ./src/ directory and run python main.py to run the project using default hyperparameters.
4. Hyperparameters can be changed from the ./src/utils.py file's get_args() function.
5. To separately evaluate the model run python evaluate_results.py from inside the ./src/ directory.

*** It is recommended to run the project in a Windows environment as all the functionalities were tested in a Windows machine. During the final testing phase, we found that the in a mac environment the system reads in images and ground truth labels in a different order than windows. This results in a mismatch between corresponding images, and the ground truth labels. ***

Directory Structure

├── data
│   └── *                   <- put the idd20k_lite.zip here before running
│── output
│   └── stacked             <- side by side images (Image, Labels (GT), Labels (Pred))
│       |───── train
│       |───── val
│       └───── test
├── results                 <- contains results about different metrics in different modes
├── documents
│   ├── docs
│   └── references
├── notebooks               <- notebooks for explorations / prototyping
└── src                     <- all source code, internal org as needed