Semantic Segmentation Example

This repository shows how to perform semantic segmentation with pytorch and tensorflow. It uses the dataset of the Kaggle Data Science Bowl 2018.

Usage

1. Install normal packages: pip install --user -r requirements.txt

2. Install Tensorflow or Pytorch

3. Download and prepare data: python data.py

4. Start jupyter notebook and use segment_tf or segment_pytorch

Different Data Set

In order to use the notebooks with different data create the folders images and masks and put the images and labels inside with the following naming convention:

• image: images/{idx}.png
• mask: masks/{idx}_mask.png