Skip to content

Segmenting roofs using satellite imagery. Efficient data loading pipeline leveraging zarr. Unique sampling strategy for faster training.

Notifications You must be signed in to change notification settings

j-sripad/Roof_segmentation

Repository files navigation

Roof Segmentation from Aerial Imagery

This repository contains code for project "Roof Segmentation from Aerial Imagery", submitted as final project for CSCI-B657(Computer Vision) course.

Team members :

  • Sripad Joshi (joshisri)
  • Srimanth Agastyraju (sragas)
  • Himani Shah (shahhi)
  • Karan Acharya (karachar)

Dataset

Aerial Imagery for Roof Segmentation

Usage details

Install dependencies

pip3 install -r roof_segmentation.txt

Run the notebooks in the following order

  • Training.ipynb : Contains data preprocessing and training code.
  • Inference.ipynb : Contains code to perform inference on the test data, calculate dice score, and overlay prediction masks on the image.
  • metrics.ipynb : This notebook plots loss curves and dice score curves for training and validation set.

Some parts of codes in this repository are inspired from a previous project one of our team members (joshisri: Sripad Joshi) worked on. The said project was his personal project and was not submitted as any assignment/project to any of the courses at Indiana University or elsewhere. It can be found at - https://github.com/j-sripad/colon_crypt_segmentation .

About

Segmenting roofs using satellite imagery. Efficient data loading pipeline leveraging zarr. Unique sampling strategy for faster training.

Resources

Stars

Watchers

Forks

Releases

No releases published

Packages

No packages published

Contributors 4

  •  
  •  
  •  
  •