stacked-unets

Getting Started

To view the code, please follow these steps:

Clone the repository by running the following command in your terminal:

git clone https://github.com/AkerkeKesha/stacked-unets.git

Navigate to the cloned repository directory:

cd stacked-unets

Create a new virtual environment using the following command:

python3 -m venv env

This will create a new folder named env in your project directory, which will contain a Python environment with its own installed packages and interpreter.

Activate the virtual environment by running the following command:

source env/bin/activate

This will activate the virtual environment and change the terminal prompt to indicate that you are now working in the virtual environment.

Once the virtual environment is activated, install the required packages using the requirements.txt file with the following command:

pip install -r requirements.txt

This will install all the packages listed in the requirements.txt file in the virtual environment. You can now edit the code as needed while working within the virtual environment. When you're done working on the project, you can deactivate the virtual environment by running the following command:

deactivate

This will return the terminal prompt to its default state, indicating that you are no longer working in the virtual environment.

Data Preparation

This project uses the ETCI 2021 Flood Detection dataset.

Steps to Download the ETCI Dataset

1. Navigate to the Kaggle Website: ETCI 2021 Flood Detection Data Card to access the dataset.

2. Register/Log in: Follow the prompts to register or log into Kaggle

3. Download: After successful login, navigate to the "Download" section and download the required dataset files.

4. Unzip the Dataset: Unzip the downloaded files into a folder.

Dataset Placement

After downloading and unzipping the dataset, move it into the dataset/ folder in this repository.

# Move the data to the dataset/ directory
mv path/to/unzipped/data/* path/to/your/repo/dataset/

Repository Structure

project-root/
|-- dataset/  # data
|-- notebooks/ # notebooks for EDA, analysis, visualizations, etc
|-- output/ # output files, such as predicted masks, plots, etc
|-- src/ # source code for train, predict, analyze, etc
|-- tests/ # py tests
|-- .gitignore # untracked files to ignore
|-- config.py # configuration file for dataset name, lr and epochs, etc
|-- gcolab_etci_flood_semantic_segmentation.ipynb # Example notebook for etci_flood in golab
|-- gcolab_spacenet6_semantic_segmentation.ipynb 
|-- README.md
|-- requirements.txt # python packages
|-- set_env.sh # Environment variables to set dataset, etc

Explore notebooks

Virtual environment can be installed as kernel in jupyter notebook

python3 -m ipykernel install --user --name=env

Now, jupyter notebooks are viewed by:

jupyter notebook

Google Colab for Training

If you don't have the computational resources to train the model locally, you can use Google Colab for training.

Initial Setup

1. Upload the Notebook: Upload gcolab_etci_flood_semantic_segmentation.ipynb to Google Colab.

2. Upload the Data:

   • Compress the dataset/ directory to dataset.zip
   • Upload dataset.zip to Google Drive.

3. Mount: Mount the Google drive in order to copy data into Colab space

Setting Environment Variables

Set up some environment variables before running the experiment. Create a code cell in your Colab notebook and paste the following:

import os

os.environ['DATASET'] = 'etci'
os.environ['ENVIRONMENT'] = 'colab'  # Set environment to 'colab'
os.environ['STOP'] = 'yes'
os.environ['OUTPUT_TYPE'] = 'softmax_prob'

Alternatively, have set_env.sh uploaded, then run:

source set_env.sh

Running the Experiment

After env variables setting up, run the experiment using the run_experiments function. Modify the parameters accordingly.

run_experiments(runs=1, n_levels=2, max_data_points=1000)

runs: The number of times to run the experiment. n_levels: The number of levels for stacking. max_data_points: The number of data points to use (useful for quick testing).