Skip to content
Learning to Steer by Mimicking Features from Heterogeneous Auxiliary Networks (AAAI 2019, oral)
Python Cuda C Shell
Branch: master
Clone or download
Fetching latest commit…
Cannot retrieve the latest commit at this time.
Permalink
Type Name Latest commit message Commit time
Failed to load latest commit information.
demo_video Add files via upload Nov 10, 2018
semantic-segmentation Update file Dec 12, 2018
steering-control Clean codes May 16, 2019
tools Remove useless codes May 16, 2019
LICENSE Create LICENSE Oct 12, 2018
README.md Provide the pre-trained model May 16, 2019

README.md

Codes for "Learning to Steer by Mimicking Features from Heterogeneous Auxiliary Networks".

Besides, our project page is now available at FM-Net.

Demo video

  • Performance of auxiliary networks on unseen target data:

  • Performance of FM-Net:

Content:

Installations

conda create -n tensorflow_gpu pip python=2.7
source activate tensorflow_gpu
pip install --upgrade tensorflow-gpu==1.4
conda install pytorch torchvision -c pytorch

Datasets

Udacity

The whole dataset is available at Udacity.

Comma-ai

The whole dataset is available at Comma-ai.

BDD100K

The whole dataset is available at BDD100K.

Semantic-Segmentation

FCN (mIoU 71.03%)

cd semantic-segmentation
python3 main.py VOCAug FCN train val --lr 0.01 --gpus 0 1 2 3 4 5 6 7 --npb

PSPNet

python3 train_pspnet.py VOCAug PSPNet train val --lr 0.01 --gpus 0 1 2 3 4 5 6 7 --npb --test_size 473

Note that you can use the code to train models (e.g., PSPNet, SegNet and FCN) in Cityscape.

Steering-Control

Test

cd steering-control
CUDA_VISIBLE_DEVICES="0" python 3d_resnet_lstm.py

Note that you need to read 3d_resnet_lstm.py and options.py carefully and modify the path accordingly. Note that current setting is used for Udacity dataset. To run the codes for Comma.ai dataset, please refer to Comma-ai and our paper to modify several parameters.

Train

CUDA_VISIBLE_DEVICES="0" python 3d_resnet_lstm.py --flag train

Note that the ImageNet pre-trained model is available here.

Performance

  1. Udacity testing set:
Model MAE RMSE
3D CNN 2.5598 3.6646
3D CNN + LSTM 1.8612 2.7167
3D ResNet (ours) 1.9167 2.8532
3D ResNet + LSTM (ours) 1.7147 2.4899
FM-Net (ours) 1.6236 2.3549
  1. Comma-ai testing set:
Model MAE RMSE
3D CNN 1.7539 2.7316
3D CNN + LSTM 1.4716 1.8397
3D ResNet (ours) 1.5427 2.4288
3D ResNet + LSTM (ours) 0.7989 1.1519
FM-Net (ours) 0.7048 0.9831
  1. BDD100K testing set:
Model Accuracy
FCN + LSTM 82.03%
3D CNN + LSTM 82.94%
3D ResNet + LSTM (ours) 83.69%
FM-Net (ours) 85.03%

Others

Citation

If you use the codes, please cite the following publications:

@article{hou2018learning,
  title={Learning to Steer by Mimicking Features from Heterogeneous Auxiliary Networks},
  author={Hou, Yuenan and Ma, Zheng and Liu, Chunxiao and Loy, Chen Change},
  journal={arXiv preprint arXiv:1811.02759},
  year={2018}
}

Acknowledgement

This repo is built upon Udacity.

Contact

If you have any problems in reproducing the results, just raise an issue in this repo.

To-Do List:

  • Release codes for steering control

  • Attach original experimental results

  • Clean all codes, make them readable and reproducable

  • Release codes for BDD100K dataset

You can’t perform that action at this time.