DASGIL: Domain Adaptation for Semantic and Geometric-aware Image-based Localization

This is our Pytorch implementation for DASGIL (paper) by Hanjiang Hu, Zhijian Qiao and Ming Cheng. The work has been published in IEEE Transactions on Image Processing (TIP).

<img src='img/overview.png' align="center" width=666 alt="Text alternative when image is not available">

Prerequisites

• Linux or macOS
• Python 3
• CPU or NVIDIA GPU + CUDA CuDNN

Getting Started

Installation

• Install requisite Python libraries.

pip install -r requirements.txt

• Clone this repo:

git clone https://github.com/HanjiangHu/DASGIL.git

Training

KITTI and Virtual KITTI 2 dataset are used to train the model, while Extended CMU-Seasons dataset is used to test. The datasets involved in this paper are well organized HERE. Please uncompress it under the root path. Our pretrained models with FD and CD are found HERE. Please uncompress it under ./checkpoints.

• Training on KITTI and Virtual KITTI Dataset (take FD model as an example, specify --dis_type CD to train and test CD model):

python train.py --name DASGIL_FD

• Fine-tune the pretrained model:

python train.py --name DASGIL_FD --continue_train --which_epoch 5

Testing

• Testing on the Extended CMU-Seasons Dataset:

python test.py --name DASGIL_FD --which_epoch 5

Results

The test results will be saved to ./output. The txt results will be merged into a single txt file for all the slices and submitted to the official benchmark website.

Our DASGIL-FD results and DASGIL-CD results could be found on the Extended CMU-Seasons benchmark website.

Other Details

• See ./options/train_options.py for training-specific flags, ./options/test_options.py for test-specific flags, and ./options/base_options.py for all common flags.
• CPU/GPU (default --gpu_ids 0): set--gpu_ids -1 to use CPU mode (NOT recommended). Currently multi-GPU training is not supported.

If you use this code in your own work, please cite:

H. Hu, Zhijian Qiao, M. Cheng, Z. Liu and H. Wang ”DASGIL: Domain Adaptation for Semantic and Geometric-aware Image-based Localization”,

@ARTICLE{hu2020dasgil,
  author={H. {Hu} and Z. {Qiao} and M. {Cheng} and Z. {Liu} and H. {Wang}},
  journal={IEEE Transactions on Image Processing}, 
  title={DASGIL: Domain Adaptation for Semantic and Geometric-Aware Image-Based Localization}, 
  year={2021},
  volume={30},
  number={},
  pages={1342-1353},
  doi={10.1109/TIP.2020.3043875}}