Wild-Places: A Large-Scale Dataset for Lidar Place Recognition in Unstructured Natural Environments

Website | Paper | Data Download Portal

This repository contains the code implementation used in the paper Wild-Places: A Large-Scale Dataset for Lidar Place Recognition in Unstructured Natural Environments, which has been accepted for publication at ICRA2023.

If you find this dataset helpful for your research, please cite our paper using the following reference:

@inproceedings{2023wildplaces,
  title={Wild-Places: A Large-Scale Dataset for Lidar Place Recognition in Unstructured Natural Environments},
  author={Knights, Joshua and Vidanapathirana, Kavisha and Ramezani, Milad and Sridharan, Sridha and Fookes, Clinton and Moghadam, Peyman},
  year={2023},
  booktitle={IEEE International Conference on Robotics and Automation (ICRA)},
  eprint={arXiv preprint arXiv:2211.12732}
}

Contents

1. Updates
2. Download Instructions
3. Scripts
   • Environment
   • Loading Point Clouds
   • Generating Training & Testing Splits
     • Training
     • Testing
   • Evaluation
     • Inter-Run
     • Intra-Run
4. Thanks

1. Updates

• 21/10/2022 Wild-Places v1.0 Uploaded
• 17/01/2023 Wild-Places is accepted to ICRA2023!
• 20/01/2023 Wild-Places v2.0 Uploaded. This update to the dataset integrates GPS into the SLAM solution to alleviate vertical drift in the larger loops of the traversal in both environments. NOTE: Sequence K-04 is currently unavailable for v2.0 due to a failed loop closure in the ground truth. We are currently working on remedying this, and will release the sequence as soon this issue is rectified.

2. Download Instructions

Our dataset can be downloaded through The CSIRO Data Access Portal. Detailed instructions for downloading the dataset can be found in the README file provided on the data access portal page.

3. Scripts

3.1 Environment

To create a python environment to use the scripts in this repository run the following command:

conda env create -f scripts/Wild-Places.yaml -n Wild-Places

3.2 Loading Point Clouds

A code snippet to load a pointcloud file from our dataset can be found in eval/load_pointcloud.py

3.2 Generating Training & Testing Splits

In this repository we provide several scripts for partitioning our dataset into splits for training and evaluation.
The output of these scripts are pickle files containing training and evaluation splits in a format compatible with existing repositories such as PointNetVLAD, MinkLoc3D(v2), TransLoc3D and PPT.

Training

To generate the training splits run the following command:

python scripts/generate_splits/training_sets.py --dataset_root $_PATH_TO_DATASET --save_folder --$_SAVE_FOLDER_PATH

Where $_PATH_TO_DATASET is the path to the downloaded dataset, and $_SAVE_FOLDER_PATH is the path to the directory where the generated files will be saved.

Testing

To generate the testing splits run the following command:

python scripts/generate_splits/testing_sets.py --dataset_root $_PATH_TO_DATASET --save_folder --$_SAVE_FOLDER_PATH

This script will generate seperate testing pickles for the inter-run and intra-run evaluation modes on each environment. The inter-run pickles will produce query and database files for each testing environment, while the intra-run pickles will produce a seperate training pickle for each individual point cloud sequence.

3.3 Evaluation

We provide evaluation scripts for both inter and intra-run evaluation on our dataset.

• For example, to evaluate the released checkpoint of LoGG3D-Net (LoGG3D-Net.pth), use the utility function provided in scripts/eval/logg3d/logg3d_utils.py to generate global descriptors for database_features and query_features. Then, use the evaluation instructions provided below:

Inter-run Evaluation

To perform inter-run evaluation on the Wild-Places dataset, run the following command:

python eval/inter-sequence.py \
    --queries $_PATH_TO_QUERIES_PICKLES \
    --databases $_PATH_TO_DATABASES_PICKLES \
    --query_features $_PATH_TO_QUERY_FEATURES \ 
    --database_features $_PATH_TO_DATABASE_FEATURES \
    --location_names $_LOCATION_NAMES \

Where:

• $_PATH_TO_QUERIES_PICKLES is a string pointing to the location of the generated query set pickle for an environment
• $_PATH_TO_DATABASES_PICKLES is a string pointing to the location of the generated database set pickle for an environment
• $_PATH_TO_QUERY_FEATURES is a string pointing towards a pickle file containing the query set features to be used in evaluation. These features should be a list of Nxd numpy arrays or tensors, where N is the number of point cloud frames in the query set of each sequence in the environment.
• $_PATH_TO_DATABASE_FEATURES is a string pointing towards a pickle file containing the database set features to be used in evaluation. These features should be a list of Nxd numpy arrays or tensors, where N is the number of point cloud frames in the database set of each sequence in the environment.
• $_LOCATION_NAMES is a string containing the name of the environment being evaluated

Intra-run Evaluation

To perform intra-run evaluation on the Wild-Places dataset, run the following command:

python eval/intra-sequence.py \
    --databases $_PATH_TO_DATABASES_PICKLES \
    --database_features $_PATH_TO_DATABASE_FEATURES \
    --run_names $_LOCATION_NAMES \

Where:

• $_PATH_TO_DATABASES_PICKLES is a string pointing to the location of the generated database set pickle for a single sequence
• $_PATH_TO_DATABASE_FEATURES is a string pointing towards a pickle file containing the run features to be used in evaluation. These features should be a single Nxd numpy array or tensor, where N is the number of point cloud frames in that sequence
• $_LOCATION_NAMES is a string containing the name of the sequence being evaluated

4. Thanks

Special thanks to the authors of the PointNetVLAD and MinkLoc3D, whose excellent code was used as a basis for the generation and evaluation scripts used in this repository.