jason08/automl

AutoDeeplab / auto-deeplab / AutoML for semantic segmentation, implemented in Pytorch

AutoML for Image Segmentation and Detection

This is an open-source project of AutoML for object detection & segmentation as well as semantic segmentation.

Currently this repo contains a pytorch implementation for Auto-Deeplab.

Following the popular trend of modern CNN architectures having a two level hierarchy. Auto-Deeplab forms a dual level search space, searching for optimal network and cell architecture.

Auto-Deeplab acheives a better performance while minimizing the size of the final model.

Architecture Search Performance

From the auto-deeplab paper	Ours

For full-sized model, leave parameters to their default setting

Training Proceedure

All together there are 3 stages:

1. Architecture Search - Here you will train one large relaxed architecture that is meant to represent many discreet smaller architectures woven together.

2. Decode - Once you've finished the architecture search, load your large relaxed architecture and decode it to find your optimal architecture.

3. Re-train - Once you have a decoded and poses a final description of your optimal model, use it to build and train your new optimal model

Architecture Search

Begin Architecture Search

Start Training

CUDA_VISIBLE_DEVICES=0 python train_autodeeplab.py --dataset cityscapes

Resume Training

CUDA_VISIBLE_DEVICES=0 python train_autodeeplab.py --dataset cityscapes --resume /AutoDeeplabpath/checkpoint.pth.tar

Multi-GPU Training

CUDA_VISIBLE_DEVICES=0,1 python train_autodeeplab.py --dataset cityscapes --batch_size 2

Load, Decode and Re-train

Now that you're done training the search algorithm, it's time to decode the search space and find your new optimal architecture. After that just build your new model and begin training it

Load and Decode

CUDA_VISIBLE_DEVICES=0 python decode_autodeeplab.py --dataset cityscapes --resume /AutoDeeplabpath/checkpoint.pth.tar

Build and Train new model

CUDA_VISIBLE_DEVICES=0 python train_new_model.py --dataset cityscapes --saved_arch_path /AutoDeeplabpathtosaveddecodings/

Requirements

• Pytorch version 1.1

• Python 3

• tensorboardX

• torchvision

• pycocotools

• tqdm

• numpy

• pandas

References

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

[2] : pytorch-deeplab-xception

[3] : Some code for the project was taken from here