/Semantic-Segmentation

Implementation of various models for Semantic Segmentation in pytorch

Primary LanguagePython

Semantic-Segmentation

Implementation of various models for Semantic Segmentation.

Currently supported segmentation models

  • FCN
  • Segnet_VGG16
  • Segnet with different pretrained networks

Installation

pip install -r requirements.txt

Data

Currently supported loaders are only VOC PASCAL.

You can find all data here:

PASCAL VOC

SBD

Config

Dataloader requires path to data. This can be provided through config.yaml.

Training

You can for example run - batch of size 6 takes approximately 9GB of gpu memory

python train.py --n_epoch 30 --batch_size 6

Credits

This repo was inspired by various other repositories and some parts of code come directly from them. So I list them all here: (Thank you)

https://github.com/bodokaiser/piwise/tree/master/piwise

https://github.com/meetshah1995/pytorch-semseg