choyaa/PiLoT-data

Dataset Card for PiLoT Dataset

We provide two releases of the PiLoT dataset:

Release	Resolution	Size	Access
Training Pack	512×512 (compressed)	~343 GB	✅ Hugging Face — Train/ + Validation/
Full Dataset	1600×1200 (original)	~6.5 TB	🔴 Baidu Netdisk / 🔵 Google Drive — see table below

Trajectory Card

• Ref-Query Pairs: Each sub-dataset includes a Reference (clear weather) and a Query (with weather effects) trajectory following the exact same path.
• Data Modalities: Sequential frames are provided as *_0.png for RGB and *_1.png for Metric Depth.
• 🙋❗Note: Please download the reprojection code from this repo and run the demo first to verify the projection alignment and coordinate system.

Download — 512×512 Training Pack

Option 1: huggingface-cli

pip install -U huggingface_hub

# Log in once (needed if the dataset is gated later)
huggingface-cli login

# Download everything (~343 GB)
huggingface-cli download choyaa/PiLoT-data --repo-type dataset --local-dir ./PiLoT-data

# Or download only one split
huggingface-cli download choyaa/PiLoT-data Train --repo-type dataset --local-dir ./PiLoT-data
huggingface-cli download choyaa/PiLoT-data Validation --repo-type dataset --local-dir ./PiLoT-data

Option 2: Python

from huggingface_hub import snapshot_download

snapshot_download(
    repo_id="choyaa/PiLoT-data",
    repo_type="dataset",
    local_dir="./PiLoT-data",
    allow_patterns=["Train/*"],  # or ["Validation/*"] or omit for all
)

Extract a trajectory

unzip PiLoT-data/Train/England_seq1@200@30_50.zip -d PiLoT-data/Train/

Full-Resolution Release (1600×1200)

In addition to the Hugging Face training pack, we provide the original 1600×1200 full-resolution dataset (~6.5 TB total) via cloud drives. Download each trajectory individually from the links below.

• Images: Download from the 🔴 Baidu Netdisk / 🔵 Google Drive links in the table.
• Poses & Cameras: Download directly from this repository.

Trajectory Name	Location (City, Country)	Lat / Lng	Height	Pitch Range	🔴 Baidu Netdisk	🔵 Google Drive
England_seq1@200@30_50	Oxford, UK	51.7345 / -1.2715	200m	[30, 50]	Link	Link
England_seq2@300@0_30	Oxford, UK	51.7653 / -1.2976	300m	[0, 30]	Link	Link
England_seq3@500@30_50	Manchester, UK	53.4687 / -2.2627	500m	[30, 50]	Link	Link
England_seq4@200@0_30	Manchester, UK	53.4846 / -2.2256	200m	[0, 30]	Link	Link
England_seq5@500@30_50	Manchester, UK	53.4689 / -2.3649	500m	[30, 50]	Link	Link
England_seq6@300@30_50	Coventry, UK	52.3789 / -1.5969	300m	[30, 50]	Link	Link
England_seq8@300@30_50	London, UK	51.5401 / -0.1380	300m	[30, 50]	Link	Link
England_seq10@200@30_50	Cambridge, UK	52.1964 / 0.0950	200m	[30, 50]	Link	Link
Finnish_seq1@200@30_50	Helsinki, Finland	60.1477 / 24.9221	200m	[30, 50]	Link	Link
Finnish_seq2@300@0_30	Helsinki, Finland	60.1913 / 24.9066	300m	[0, 30]	Link	Link
Finnish_seq4@200@0_30	Espoo, Finland	60.2041 / 24.7858	200m	[0, 30]	Link	Link
Finnish_seq5@500@30_50	Turku, Finland	60.4387 / 22.2074	500m	[30, 50]	Link	Link
Finnish_seq6@300@30_50	Turku, Finland	60.4617 / 22.2642	300m	[30, 50]	Link	Link
Finnish_seq8@300@30_50	Tampere, Finland	61.5100 / 23.7607	300m	[30, 50]	Link	Link
Finnish_seq10@200@30_50	Vantaa, Finland	60.3131 / 25.0673	200m	[30, 50]	Link	Link
France_seq1@200@30_50	Paris, France	48.8460 / 2.3116	200m	[30, 50]	Link	Link
France_seq2@300@0_30	Paris, France	48.8724 / 2.2739	300m	[0, 30]	Link	Link
France_seq4@200@0_30	Roissy-en-France, France	49.0151 / 2.5308	200m	[0, 30]	Link	Link
France_seq5@500@30_50	Paris, France	48.8805 / 2.3694	500m	[30, 50]	Link	Link
France_seq8@300@30_50	Metz, France	49.1235 / 6.1455	300m	[30, 50]	Link	Link
France_seq9@500@0_30	Bordeaux, France	44.8501 / -0.5819	500m	[0, 30]	Link	Link
France_seq10@200@30_50	Marseille, France	43.3191 / 5.3817	200m	[30, 50]	Link	Link
German_seq1@200@30_50	Munich, Germany	48.1544 / 11.5428	200m	[30, 50]	Link	Link
German_seq2@300@0_30	Aachen, Germany	50.7750 / 6.0724	300m	[0, 30]	Link	Link
German_seq4@200@0_30	Munich, Germany	48.1877 / 11.5741	200m	[0, 30]	Link	Link
German_seq5@500@30_50	Frankfurt, Germany	50.0439 / 8.5277	500m	[30, 50]	Link	Link
German_seq8@300@30_50	Berlin, Germany	52.4975 / 13.3710	300m	[30, 50]	Link	Link
German_seq9@500@0_30	Frankfurt, Germany	50.1100 / 8.6641	500m	[0, 30]	Link	Link
German_seq10@200@30_50	Heidelberg, Germany	49.4041 / 8.6462	200m	[30, 50]	Link	Link
Italy_seq1@200@30_50	Milan, Italy	45.4467 / 9.2271	200m	[30, 50]	Link	Link
Italy_seq2@300@0_30	Milan, Italy	45.4910 / 9.1526	300m	[0, 30]	Link	Link
Italy_seq3@500@30_50	Rome, Italy	41.9260 / 12.4894	500m	[30, 50]	Link	Link
Italy_seq4@200@0_30	Rome, Italy	41.9161 / 12.4628	200m	[0, 30]	Link	Link
Italy_seq5@500@30_50	Bologna, Italy	44.4906 / 11.3297	500m	[30, 50]	Link	Link
Italy_seq6@300@30_50	Bologna, Italy	44.4776 / 11.3755	300m	[30, 50]	Link	Link
Italy_seq8@300@30_50	Florence, Italy	43.8488 / 11.1064	300m	[30, 50]	Link	Link
Italy_seq9@500@0_30	Venice, Italy	45.4405 / 12.3005	500m	[0, 30]	Link	Link
Netherland_seq1@500@0_30	Amsterdam, Netherlands	52.3492 / 4.8506	500m	[0, 30]	Link	Link
Netherland_seq4@200@30_50	Amsterdam, Netherlands	52.3391 / 4.8653	200m	[30, 50]	Link	Link
Netherland_seq5@300@0_30	The Hague, Netherlands	52.0709 / 4.2502	300m	[0, 30]	Link	Link
Netherland_seq7@200@0_30	Delft, Netherlands	51.9898 / 4.3614	200m	[0, 30]	Link	Link
Netherland_seq8@500@30_50	Schiphol, Netherlands	52.3203 / 4.7171	500m	[30, 50]	Link	Link
Netherland_seq9@300@30_50	Maastricht, Netherlands	50.8688 / 5.6485	300m	[30, 50]	Link	Link
Netherland_seq10@200@0_30	Tilburg, Netherlands	51.5559 / 5.0494	200m	[0, 30]	Link	Link
spain_seq1@500@0_30	Barcelona, Spain	41.3697 / 2.1916	500m	[0, 30]	Link	Link
spain_seq2@300@30_50	Barcelona, Spain	41.3544 / 2.1485	300m	[30, 50]	Link	Link
spain_seq3@300@30_50	Valencia, Spain	39.4519 / -0.3441	300m	[30, 50]	Link	Link
spain_seq4@500@30_50	Valencia, Spain	39.4604 / -0.3035	500m	[30, 50]	Link	Link
spain_seq5@200@30_50	Madrid, Spain	40.5479 / -3.7035	200m	[30, 50]	Link	Link
spain_seq6@200@0_30	Madrid, Spain	40.4163 / -3.7145	200m	[0, 30]	Link	Link
spain_seq8@500@0_30	San Sebastián, Spain	43.3251 / -1.9178	500m	[0, 30]	Link	Link
spain_seq9@300@30_50	Seville, Spain	37.3611 / -6.0668	300m	[30, 50]	Link	Link
Switzerland_seq1@500@30_50	Lausanne, Switzerland	46.5368 / 6.5305	500m	[30, 50]	Link	Link
Switzerland_seq2@200@0_30	Basel, Switzerland	47.6281 / 7.5723	200m	[0, 30]	Link	Link
Switzerland_seq3@300@30_50	Zurich, Switzerland	47.3749 / 8.5493	300m	[30, 50]	Link	Link
Switzerland_seq5@300@0_30	Lausanne, Switzerland	46.5178 / 6.5259	300m	[0, 30]	Link	Link
Switzerland_seq6@200@30_50	Lausanne, Switzerland	46.5087 / 6.5445	200m	[30, 50]	Link	Link
Switzerland_seq7@500@30_50	Geneva, Switzerland	43.3265 / -1.9552	500m	[30, 50]	Link	Link
Switzerland_seq8@300@30_50	Geneva, Switzerland	43.3251 / -1.9178	300m	[30, 50]	Link	Link
Switzerland_seq9@500@30_50	Bern, Switzerland	37.3611 / -6.0668	500m	[30, 50]	Link	Link
Switzerland_seq10@200@0_30	Bern, Switzerland	37.3685 / -6.0130	200m	[0, 30]	Link	Link
Switzerland_seq12@300@0_30	Lucerne, Switzerland	47.0303 / 8.2557	300m	[0, 30]	Link	Link
Switzerland_seq13@500@0_30	Lucerne, Switzerland	47.0536 / 8.2621	500m	[0, 30]	Link	Link
Switzerland_seq14@200@0_30	Lucerne, Switzerland	47.0897 / 8.2462	200m	[0, 30]	Link	Link
Switzerland_seq16@300@30_50	Zurich, Switzerland	47.3534 / 8.5334	300m	[30, 50]	Link	Link
Switzerland_seq17@200@0_30	Zurich, Switzerland	47.3876 / 8.5052	200m	[0, 30]	Link	Link
Switzerland_seq18@300@30_50	Zurich, Switzerland	47.3971 / 8.5259	300m	[30, 50]	Link	Link
Switzerland_seq19@500@0_30	Zurich, Switzerland	47.3894 / 8.4962	500m	[0, 30]	Link	Link
Switzerland_seq20@200@30_50	Zurich, Switzerland	47.4094 / 8.4573	200m	[30, 50]	Link	Link
Switzerland_seq21@200@30_50	Geneva, Switzerland	46.2181 / 6.1262	200m	[30, 50]	Link	Link
Switzerland_seq22@300@0_30	Geneva, Switzerland	46.2271 / 6.1392	300m	[0, 30]	Link	Link
Switzerland_seq23@500@0_30	Geneva, Switzerland	46.2002 / 6.1098	500m	[0, 30]	Link	Link
Switzerland_seq24@200@0_30	Geneva, Switzerland	46.1817 / 6.0968	200m	[0, 30]	Link	Link
Switzerland_seq25@500@30_50	Bern, Switzerland	46.9171 / 7.4149	500m	[30, 50]	Link	Link
Switzerland_seq26@300@30_50	Bern, Switzerland	46.9410 / 7.3823	300m	[30, 50]	Link	Link
Switzerland_seq27@200@0_30	Bern, Switzerland	46.9316 / 7.4421	200m	[0, 30]	Link	Link
Switzerland_seq28@300@30_50	St. Gallen, Switzerland	47.4465 / 9.4201	300m	[30, 50]	Link	Link
Switzerland_seq29@500@0_30	Winterthur, Switzerland	47.4770 / 8.7008	500m	[0, 30]	Link	Link
Switzerland_seq30@200@30_50	Winterthur, Switzerland	47.5307 / 8.7592	200m	[30, 50]	Link	Link
Switzerland_seq31@200@30_50	Winterthur, Switzerland	47.4933 / 8.7005	200m	[30, 50]	Link	Link
Switzerland_seq32@300@0_30	Winterthur, Switzerland	47.5323 / 8.6611	300m	[0, 30]	Link	Link
Switzerland_seq33@500@0_30	Fribourg, Switzerland	46.8333 / 7.1405	500m	[0, 30]	Link	Link
Switzerland_seq34@200@0_30	Fribourg, Switzerland	46.8096 / 7.1637	200m	[0, 30]	Link	Link
Switzerland_seq35@500@30_50	Neuchâtel, Switzerland	46.9911 / 6.9122	500m	[30, 50]	Link	Link
Switzerland_seq36@300@30_50	Biel, Switzerland	47.1306 / 7.2319	300m	[30, 50]	Link	Link
Switzerland_seq37@200@0_30	Biel, Switzerland	47.1235 / 7.2386	200m	[0, 30]	Link	Link
Switzerland_seq38@300@30_50	Biel, Switzerland	47.1100 / 7.2220	300m	[30, 50]	Link	Link
Switzerland_seq39@500@0_30	Lugano, Switzerland	45.9913 / 8.9424	500m	[0, 30]	Link	Link
Switzerland_seq40@200@30_50	Thun, Switzerland	46.7695 / 7.6109	200m	[30, 50]	Link	Link
Switzerland_seq43@500@0_30	Meyrin, Switzerland	46.2205 / 6.0477	500m	[0, 30]	Link	Link
Switzerland_seq44@200@0_30	Lausanne, Switzerland	46.5432 / 6.5679	200m	[0, 30]	Link	Link
Switzerland_seq45@500@30_50	Lugano, Switzerland	45.9938 / 8.9081	500m	[30, 50]	Link	Link
Switzerland_seq46@300@30_50	Schaffhausen, Switzerland	47.7004 / 8.6137	300m	[30, 50]	Link	Link
Switzerland_seq47@200@0_30	Schaffhausen, Switzerland	47.7230 / 8.6519	200m	[0, 30]	Link	Link
Switzerland_seq48@300@30_50	Schaffhausen, Switzerland	47.6934 / 8.6037	300m	[30, 50]	Link	Link
Switzerland_seq49@500@0_30	Sion, Switzerland	46.2320 / 7.3857	500m	[0, 30]	Link	Link
Switzerland_seq50@200@30_50	Zug, Switzerland	47.1853 / 8.5280	200m	[30, 50]	Link	Link
Switzerland_seq51@200@30_50	Zug, Switzerland	47.1870 / 8.4719	200m	[30, 50]	Link	Link
Switzerland_seq52@300@0_30	Zug, Switzerland	47.2009 / 8.4735	300m	[0, 30]	Link	Link
Switzerland_seq53@500@0_30	Solothurn, Switzerland	47.1969 / 7.5088	500m	[0, 30]	Link	Link
USA_seq1@200@0_30	Chicago, USA	41.8891 / -87.6396	200m	[0, 30]	Link	Link
USA_seq3@500@0_30	Las Vegas, USA	36.0927 / -115.1618	500m	[0, 30]	Link	Link
USA_seq5@500@0_30	New York, USA	40.7302 / -73.9710	500m	[0, 30]	Link	Link
USA_seq6@300@30_50	San Francisco, USA	37.7739 / -122.3873	300m	[30, 50]	Link	Link
USA_seq10@200@0_30	Washington D.C., USA	38.8798 / -76.9822	200m	[0, 30]	Link	Link

Citation

If you find this dataset useful, please cite our CVPR 2026 Highlight paper:

@misc{cheng2026pilotneuralpixelto3dregistration,
      title={PiLoT: Neural Pixel-to-3D Registration for UAV-based Ego and Target Geo-localization}, 
      author={Xiaoya Cheng and Long Wang and Yan Liu and Xinyi Liu and Hanlin Tan and Yu Liu and Maojun Zhang and Shen Yan},
      year={2026},
      eprint={2603.20778},
      archivePrefix={arXiv},
      primaryClass={cs.CV},
      url={https://arxiv.org/abs/2603.20778}, 
}

Licensing Information

Copyright (c) 2026 Saw Lab, National University of Defense Technology (NUDT).

The Dataset is provided for non-commercial research and educational purposes only.

Terms of Use:

• Attribution: You must provide appropriate credit to Saw Lab, NUDT and cite the corresponding CVPR 2026 paper in any derivative works or publications.
• Non-Commercial: Commercial use, including but not limited to selling the data or using it to train commercial models, is strictly prohibited.

---