thewontonsoup/SemanticSegmentationElectricity

A project using existing datasets to predict areas within SubSaharan Africa without Electricity using Semantic Segmentation Techniques # Settlement and Electricity Detection ## Authors ## The Data ** The following description is taken directly from the IEEE GRSS 2021 Challenge website. The IEEE GRSS 2021 ESD dataset is composed of 98 tiles of 800×800 pixels, distributed respectively across the training, validation and test sets as follows: 60, 19, and 19 tiles. Each tile includes 98 channels from the below listed satellite images. Please note that all the images have been resampled to a Ground Sampling Distance (GSD) of 10 m. Thus each tile corresponds to a 64km2 area. ## Prerequisite - Python 3.10.11: Install from Python's official website - CUDA Toolkit 11.8.0: Install from NVIDIA’s official website ## Setting Up virtual environment To make sure you download all the packages to begin this homework assignment we will utilize a Python virtual environment which is an isolated environment that allows you to run the homework with its own dependencies and libraries independent of other Python projects you may be working on. Here's how to set it up: 1. Create a virtual environment: `python3 -m venv venv` 2. Activate the virtual environment: * On macOS and Linux: `source venv/bin/activate` * On Windows: `venv\Scripts\activate` 3. (Optional) We suggest calling removing any problematic cached pip packages: `pip cache purge` 4. Install the required packages: `pip install -r requirements.txt` ## Directions Clone this repository e.g. `git clone git@github.com:cs175cv-w2024/final-project-team-tsvt.git` Please download and unzip the `dfc2021_dse_train.zip` saving the `Train` directory into the `data/raw` directory. The zip file is available at the following url. To train your own model: - Use the default values of the ESDConfig in the provided scripts/train.py or configure the provided scripts/train.yml to the desired model and parameters - Run the script by entering `python -m scripts.train` - Supported Models: `SegmentationCNN`, `UNet`, `FCNResnetTransfer`, `UNetPlusPlus …