https://github.com/shreydan/binary-semantic-segmentation

binary semantic segmentation using UNet on CamVid dataset
https://github.com/shreydan/binary-semantic-segmentation

Last synced: 8 months ago
JSON representation

binary semantic segmentation using UNet on CamVid dataset

• Host: GitHub
• URL: https://github.com/shreydan/binary-semantic-segmentation
• Owner: shreydan
• Created: 2022-12-03T13:04:18.000Z (about 3 years ago)
• Default Branch: main
• Last Pushed: 2022-12-03T14:14:31.000Z (about 3 years ago)
• Last Synced: 2024-03-15T14:16:07.416Z (almost 2 years ago)
• Language: Jupyter Notebook
• Homepage:
• Size: 25.8 MB
• Stars: 0
• Watchers: 2
• Forks: 0
• Open Issues: 0
• Metadata Files:
  • Readme: README.md

Awesome Lists containing this project

README

          
# Instance Segmentation

## Binary Segmentation

Detecting roads in car driving clips - Reaching validation accuracy of 95%

model: U-NET

the steps are mentioned briefly in the notebook.

- [notebook](./u-net-binary-instance-segmentation-pytorch.ipynb)

- [kaggle notebook](https://www.kaggle.com/code/shreydan/u-net-binary-instance-segmentation-pytorch/notebook)

Original Segmentation Dataset (CamVid):

![](https://media1.giphy.com/media/QeMYXIMpakTgJXr1Cg/giphy.gif)

Detecting Roads in the dataset:

![](https://media3.giphy.com/media/sKxAUV8WPqsm1gRGOq/giphy.gif)

![](https://media0.giphy.com/media/SPzdI0gYWvJBYVey4f/giphy.gif)

### TODO

- multiclass segmentation

- train on other datasets such as Cityscapes

- combining datasets

- better resolution

- look at other Loss functions like dice loss

- understand IoU

___