license: mit
tags:
- event-camera
- localization
- visual-place-recognition
- robotics
- image-feature-extraction
Gardens Point Event Dataset for viewpoint variant localization
This repository contains the datasets for the Gardens Point Event Dataset, consisting of two walking traverses of the same route with a severe left and right viewpoint shift.
The dataset was captured with an iniVation DAVIS346 event camera mounted to a handheld gimble with an iPhone 16 recording RGB frames and providing GPS ground truth positioning.
Dataset structure
The dataset has the following structure:
Gardens-Point-Event/
├── left/
| ├── derived/
| | ├── gps_interp.csv
| | ├── sync.json
| ├── camera_sync.csv
| ├── frames.zip
| ├── phone.jsonl
| ├── left.hdf5
| ├── session.json
|
├── right/
| ├── derived/
| | ├── gps_interp.csv
| | ├── sync.json
| ├── camera_sync.csv
| ├── frames.zip
| ├── phone.jsonl
| ├── right.hdf5
| ├── session.json
The .hdf5 files contain the raw event stream from the DAVIS346, where the derived/gps_interp.csv contains the interpolated ground truth from the camera sync. frames.zip
contains RGB .jpg images from the iPhone camera taken at 1Hz, which can be used for further alignment.
session.json contains metadata about the recordings.
If you use this dataset in your work, please cite:
@misc{hines2026eventgem,
title={EventGeM: Global-to-Local Feature Matching for Event-Based Visual Place Recognition},
author={Adam D. Hines and Gokul B. Nair and Nicolás Marticorena and Michael Milford and Tobias Fischer},
year={2026},
eprint={2603.05807},
archivePrefix={arXiv},
primaryClass={cs.CV},
url={https://arxiv.org/abs/2603.05807},
}