Updated code from our TPAMI paper.

xMUDA: Cross-Modal Unsupervised Domain Adaptation for 3D Semantic Segmentation

Official code for the paper.

Paper

xMUDA: Cross-Modal Unsupervised Domain Adaptation for 3D Semantic Segmentation
Maximilian Jaritz, Tuan-Hung Vu, Raoul de Charette, Émilie Wirbel, Patrick Pérez
Inria, valeo.ai CVPR 2020

If you find this code useful for your research, please cite our paper:

@inproceedings{jaritz2019xmuda,
	title={{xMUDA}: Cross-Modal Unsupervised Domain Adaptation for {3D} Semantic Segmentation},
	author={Jaritz, Maximilian and Vu, Tuan-Hung and de Charette, Raoul and Wirbel, Emilie and P{\'e}rez, Patrick},
	booktitle={CVPR},
	year={2020}
}

Preparation

Prerequisites

Tested with

• PyTorch 1.4
• CUDA 10.0
• Python 3.8
• SparseConvNet
• nuscenes-devkit

Installation

As 3D network we use SparseConvNet. It requires to use CUDA 10.0 (it did not work with 10.1 when we tried). We advise to create a new conda environment for installation. PyTorch and CUDA can be installed, and SparseConvNet installed/compiled as follows:

$ conda install pytorch torchvision cudatoolkit=10.0 -c pytorch
$ pip install --upgrade git+https://github.com/facebookresearch/SparseConvNet.git

Clone this repository and install it with pip. It will automatically install the nuscenes-devkit as a dependency.

$ git clone https://github.com/valeoai/xmuda.git
$ cd xmuda
$ pip install -ve .

The -e option means that you can edit the code on the fly.

Datasets

NuScenes

Please download the Full dataset (v1.0) from the NuScenes website and extract it.

You need to perform preprocessing to generate the data for xMUDA first. The preprocessing subsamples the 360° LiDAR point cloud to only keep the points that project into the front camera image. It also generates the point-wise segmentation labels using the 3D objects by checking which points lie inside the 3D boxes. All information will be stored in a pickle file (except the images which will be read frame by frame by the dataloader during training).

Please edit the script xmuda/data/nuscenes/preprocess.py as follows and then run it.

• root_dir should point to the root directory of the NuScenes dataset
• out_dir should point to the desired output directory to store the pickle files

A2D2

Please download the Semantic Segmentation dataset and Sensor Configuration from the Audi website or directly use wget and the following links, then extract.

$ wget https://aev-autonomous-driving-dataset.s3.eu-central-1.amazonaws.com/camera_lidar_semantic.tar
$ wget https://aev-autonomous-driving-dataset.s3.eu-central-1.amazonaws.com/cams_lidars.json

The dataset directory should have this basic structure:

a2d2                                   % A2D2 dataset root
 ├── 20180807_145028
 ├── 20180810_142822
 ├── ...
 ├── cams_lidars.json
 └── class_list.json

For preprocessing, we undistort the images and store them separately as .png files. Similar to NuScenes preprocessing, we save all points that project into the front camera image as well as the segmentation labels to a pickle file.

Please edit the script xmuda/data/a2d2/preprocess.py as follows and then run it.

• root_dir should point to the root directory of the A2D2 dataset
• out_dir should point to the desired output directory to store the undistorted images and pickle files. It should be set differently than the root_dir to prevent overwriting of images.

SemanticKITTI

Please download the files from the SemanticKITTI website and additionally the color data from the Kitti Odometry website. Extract everything into the same folder.

Similar to NuScenes preprocessing, we save all points that project into the front camera image as well as the segmentation labels to a pickle file.

Please edit the script xmuda/data/semantic_kitti/preprocess.py as follows and then run it.

• root_dir should point to the root directory of the SemanticKITTI dataset
• out_dir should point to the desired output directory to store the pickle files

Training

xMUDA

You can run the training with

$ cd <root dir of this repo>
$ python xmuda/train_xmuda.py --cfg=configs/nuscenes/usa_singapore/xmuda.yaml

The output will be written to /home/<user>/workspace/outputs/xmuda/<config_path> by default. The OUTPUT_DIR can be modified in the config file in (e.g. configs/nuscenes/usa_singapore/xmuda.yaml) or optionally at run time in the command line (dominates over config file). Note that @ in the following example will be automatically replaced with the config path, i.e. with nuscenes/usa_singapore/xmuda.

$ python xmuda/train_xmuda.py --cfg=configs/nuscenes/usa_singapore/xmuda.yaml OUTPUT_DIR path/to/output/directory/@

You can start the trainings on the other UDA scenarios (Day/Night and A2D2/SemanticKITTI) analogously:

$ python xmuda/train_xmuda.py --cfg=configs/nuscenes/day_night/xmuda.yaml
$ python xmuda/train_xmuda.py --cfg=configs/a2d2_semantic_kitti/xmuda.yaml

xMUDA_PL

After having trained the xMUDA model, generate the pseudo-labels as follows:

$ python xmuda/test.py --cfg=configs/nuscenes/usa_singapore/xmuda.yaml --pselab @/model_2d_100000.pth @/model_3d_100000.pth DATASET_TARGET.TEST "('train_singapore',)"

Note that we use the last model at 100,000 steps to exclude supervision from the validation set by picking the best weights. The pseudo labels and maximum probabilities are saved as .npy file.

Please edit the pselab_paths in the config file, e.g. configs/nuscenes/usa_singapore/xmuda_pl.yaml, to match your path of the generated pseudo-labels.

Then start the training. The pseudo-label refinement (discard less confident pseudo-labels) is done when the dataloader is initialized.

$ python xmuda/train_xmuda.py --cfg=configs/nuscenes/usa_singapore/xmuda_pl.yaml

You can start the trainings on the other UDA scenarios (Day/Night and A2D2/SemanticKITTI) analogously:

$ python xmuda/test.py --cfg=configs/nuscenes/day_night/xmuda.yaml --pselab @/model_2d_100000.pth @/model_3d_100000.pth DATASET_TARGET.TEST "('train_night',)"
$ python xmuda/train_xmuda.py --cfg=configs/nuscenes/day_night/xmuda_pl.yaml

# use batch size 1, because of different image sizes Kitti
$ python xmuda/test.py --cfg=configs/a2d2_semantic_kitti/xmuda.yaml --pselab @/model_2d_100000.pth @/model_3d_100000.pth DATASET_TARGET.TEST "('train',)" VAL.BATCH_SIZE 1
$ python xmuda/train_xmuda.py --cfg=configs/a2d2_semantic_kitti/xmuda_pl.yaml

Baseline

Train the baselines (only on source) with:

$ python xmuda/train_baseline.py --cfg=configs/nuscenes/usa_singapore/baseline.yaml
$ python xmuda/train_baseline.py --cfg=configs/nuscenes/day_night/baseline.yaml
$ python xmuda/train_baseline.py --cfg=configs/a2d2_semantic_kitti/baseline.yaml

Testing

You can provide which checkpoints you want to use for testing. We used the ones that performed best on the validation set during training (the best val iteration for 2D and 3D is shown at the end of each training). Note that @ will be replaced by the output directory for that config file. For example:

$ cd <root dir of this repo>
$ python xmuda/test.py --cfg=configs/nuscenes/usa_singapore/xmuda.yaml @/model_2d_065000.pth @/model_3d_095000.pth

You can also provide an absolute path without @.

Model Zoo

You can download the models with the scores below from this Google drive folder.

Method	USA/Singapore 2D	USA/Singapore 3D	Day/Night 2D	Day/Night 3D	A2D2/Sem.KITTI 2D	A2D2/Sem.KITTI 3D
Baseline (source only)	53.4	46.5	42.2	41.2	34.2*	35.9*
xMUDA	59.3	52.0	46.2	44.2	38.3*	46.0*
xMUDAPL	61.1	54.1	47.1	46.7	41.2*	49.8*

^* Slight differences from the paper on A2D2/Sem.KITTI: Now we use class weights computed on source. In the paper, we falsely computed class weights on the target domain.

Acknowledgements

Note that this code borrows from the MVPNet repo.

License

xMUDA is released under the Apache 2.0 license.