sanimorla/toronto-3d

A Large-scale Mobile LiDAR Dataset for Semantic Segmentation of Urban Roadways

Toronto-3D: A Large-scale Mobile LiDAR Dataset for Semantic Segmentation of Urban Roadways

[Paper] [Download]

Toronto-3D is a large-scale urban outdoor point cloud dataset acquired by an MLS system in Toronto, Canada for semantic segmentation. This dataset covers approximately 1 km of road and consists of about 78.3 million points. Point clouds has 10 attributes and classified in 8 labelled object classes. There is a data preparation tip to handle UTM coordinates to avoid problems. There are also some known issues.

Details on the dataset can be found at CVPRW2020. Revisions on the labels will lead to different results from the published paper, and updated results will be updated here.

If you have questions, or any suggestions to help us improve the dataset, please contact Weikai Tan.

---

Download

Dataset can be downloaded at OneDrive or 百度网盘(提取码：aewp). Check Changelog for changes.

Toronto-3D belongs to Mobile Sensing and Geodata Science Lab, University of Waterloo. Toronto-3D is distributed under the CC BY-NC 4.0 License

Citation

Please consider citing our work:

@inproceedings{tan2020toronto3d,
    title={Toronto-3D: A large-scale mobile lidar dataset for semantic segmentation of urban roadways},
    author={Tan, Weikai and Qin, Nannan and Ma, Lingfei and Li, Ying and Du, Jing and Cai, Guorong and Yang, Ke and Li, Jonathan},
    booktitle={Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition Workshops},
    pages={202--203},
    year={2020}
}

Acknowledgements

Teledyne Optech is acknowledged for providing mobile LiDAR point cloud data collected by Maverick. Thanks Jing Du and Dr. Guorong Cai from Jimei University for point cloud labelling.

Thanks Intel ISL for including our dataset in the Open3D-ML 3D Machine Learning module.

---

Semantic segmentation results with point coordinates

More results to be added

Method	OA	mIoU	Road	Road marking	Natural	Building	Utility line	Pole	Car	Fence
PointNet++	84.88%	41.81%	89.27%	0.00%	69.06%	54.16%	43.78%	23.30%	52.00%	2.95%
PointNet++ *	91.66%	58.01%	92.71%	7.68%	84.30%	81.83%	67.44%	63.30%	60.92%	5.92%
PointNet++ MSG	92.56%	59.47%	92.90%	0.00%	86.13%	82.15%	60.96%	62.81%	76.41%	14.43%
DGCNN	94.24%	61.79%	93.88%	0.00%	91.25%	80.39%	62.40%	62.32%	88.26%	15.81%
KPFCNN	95.39%	69.11%	94.62%	0.06%	96.07%	91.51%	87.68%	81.56%	85.66%	15.72%
MS-PCNN	90.03%	65.89%	93.84%	3.83%	93.46%	82.59%	67.80%	71.95%	91.12%	22.50%
TGNet	94.08%	61.34%	93.54%	0.00%	90.83%	81.57%	65.26%	62.98%	88.73%	7.85%
MS-TGNet	95.71%	70.50%	94.41%	17.19%	95.72%	88.83%	76.01%	73.97%	94.24%	23.64%
RandLA-Net	88.43%	74.27%	87.43%	22.04%	96.36%	92.69%	85.93%	75.50%	86.60%	47.64%

* use same radii and k as TGNet

---

Point cloud attributes

• XYZ
• RGB
• Intensity
• GPS time
• Scan angle rank

Classes

• Road (label 1)
• Road marking (label 2)
• Natural (label 3)
• Building (label 4)
• Utility line (label 5)
• Pole (label 6)
• Car (label 7)
• Fence (label 8)
• unclassified (label 0)

---

Data preparation tip

The XY coordinates are stored in UTM format. The Y coordinate may exceed decimal digits in float type commonly used in point cloud processing algorithms. Directly read and process the coordinates could result in loss of detail and wrong geometric features.

I set a UTM_OFFSET = [627285, 4841948, 0] to subtract from the raw coordinates. You may use any other numbers to reduce number of digits.

Example of potential issues during grid_subsampling operation used in KPConv and RandLA-Net: both subsampled to grid size 6cm

without offset	with offset

---

Known issues

1. Point RGB assignments on taller vehicles.

2. Point RGB artifact assignments on moving vehicles.

3. Point acquisition on moving vehicles.

---

Changelog

• [2020-04-23] Uploaded newest version. Fixed some labelling errors. Major revision on cars.

• [2020-03-22] Uploaded newest version.