Skip to content

Road Segmentation from Satelite images using custom Unet model

Notifications You must be signed in to change notification settings

sunnynevarekar/Road-Segmentation

Folders and files

NameName
Last commit message
Last commit date

Latest commit

 

History

13 Commits
 
 
 
 
 
 
 
 
 
 
 
 
 
 

Repository files navigation

Road Segmentation from Satelite images using a custom Unet model

This repository contains code for training and evaluating a custom Unet model to segment roads from satelite images.

Dataset

The dataset can be downloaded from this kaggle link. Some of the images in training set does not have corresponding masks. The training code filters out those images. All the images are of size 1500x1500. Code for data analysis is in this notebook.

Model Architecture

main.py file contains code for training and testing.

E.g.

For training

python main.py train --train_dir=train_dir --model_save_path=path_to_save_model --batch_size=batch_size --epochs=num_epochs

To check all the parameters for training

python main.py train --help

For testing

python main.py test --test_dir=test_dir --checkpoint_path=trained_model_checkpoint_path --save_preds

To check all the parameters for testing

python main.py test --help

About

Road Segmentation from Satelite images using custom Unet model

Topics

Resources

Stars

Watchers

Forks

Releases

No releases published

Packages

No packages published