Awesome

seg-eval

seg-eval is a Python library that provides tools for evaluating semantic segmentation models.

This library is associated to the paper Guidelines to Compare Semantic Segmentation Maps at Different Resolutions.

What's inside

• Pixel-wise semantic segmentation metrics.
• Utilities to generate evaluation regions and to analyze segmentation results within them.
• Functions to qualitatively analyze segmentation results and to generate figures for academic papers.

Installation

Since the package has not been converted into a pip installable package yet, you can install it by cloning this repository.

git clone https://github.com/itracasa/seg-eval.git

Example notebooks

• 1. metrics module: Pixel-wise semantic segmentation metrics.
• 2. region_metrics module: Utilities to generate evaluation regions and to analyze segmentation results within them.

The visualization module is used in both of the notebooks.

Citation