Statement

Thanks for your attention to the project.Now,the project can't reproduce the results.It just provides some ideas to reproduce the results.If you have some good ideas, both merging and discussing are welcome.

AutoDeeplab

This is an implementation of Auto-DeepLab using Pytorch.

Environment

The implementation needs the following dependencies:

Python = 3.7
Pytorch = 0.4
TensorboardX

Other basic dependencies like matplotlib, tqdm ... are also needed.

Installation

First, clone the repository

git clone https://github.com/MenghaoGuo/AutoDeeplab.git

Then

cd AutoDeeplab

Train

The dataloader module is built on this repo

If you want to train this model on different datasets, you need to edit --dataset parameter and then:

bash train_voc.sh

Reference

[1] : Auto-DeepLab: Hierarchical Neural Architecture Search for Semantic Image Segmentation