Academy
Docs
- New dataset
- New organization
Sign in

The point cloud segmentation interface is currently only available at webgpu.segments.ai.

segments /

nuscenes-3d-segmentation

Public

Point Cloud Seq

Segmentation

Overview
Samples
Issues
Releases

Label set

ground-truth

Create a new label set

A label set is a collection of labels. Create a new label set if you want to upload model predictions and use features such as model-assisted labeling.

Name

Description (optional)

Set up the SDK

From the command line, install the Segments.ai Python SDK.

pip install --upgrade segments-ai

Initialize the Segments client using your API key.

from segments import SegmentsClient

...api_key = ""

client = SegmentsClient(api_key)

Copy the code block using the button to obtain your full API key.

Create a new label set

Create a new label set with client.add_labelset(). You need to specify the name and (optionally) the description of the new label set.

dataset_identifier = "segments/nuscenes-3d-segmentation"
name = "model-predictions"

client.add_labelset(dataset_identifier, name)

Search syntax

This is an example dataset

For best performance, create a new dataset and choose a suitable data category.

Upload errors

Name	Label status	Last updated
scene-0061_3_frames	Prelabeled	3 years ago
scene-0103	Prelabeled	3 years ago
scene-0553	Prelabeled	3 years ago
scene-0655	Prelabeled	3 years ago
scene-0757	Prelabeled	8 months ago
scene-0796	Prelabeled	3 years ago
scene-0916	Prelabeled	3 years ago
scene-1077	Prelabeled	3 years ago
scene-1094	Prelabeled	3 years ago

Add samples

Click to select files
or drag and drop files

.pcd, .bin, .ply

Set up the SDK

From the command line, install the Segments.ai Python SDK.

pip install --upgrade segments-ai

Initialize the Segments client using your API key.

from segments import SegmentsClient

...api_key = ""

client = SegmentsClient(api_key)

Copy the code block using the button to obtain your full API key.

Upload a file

Use client.upload_asset() to upload a local file to Segments.ai's asset storage service.

filename = "ca9a282c9e77460f8360f564131a8af5_nuscenes.bin"

with open(f"path/to/{filename}", "rb") as f:
    asset = client.upload_asset(f, filename=filename)

point_cloud_url = asset.url

Add a sample

Add a sample with client.add_sample(). For each sample, specify a name and the required sample attributes. If you do not have URLs for your assets, upload the assets to Segments.ai first.

dataset = "segments/nuscenes-3d-segmentation"

name = "sample_point_cloud_sequence"

pc_urls = [
    "https://segmentsai-prod.s3.eu-west-2.amazonaws.com/assets/admin-tobias/73b62f59-cb43-4625-aeb0-6e64ea42b28a.pcd",
    "https://segmentsai-prod.s3.eu-west-2.amazonaws.com/assets/admin-tobias/73b62f59-cb43-4625-aeb0-6e64ea42b28a.pcd",
]

frames = [
    {
        "pcd": {"url": pc_urls[0], "type": "pcd"},
        "name": "sample_point_cloud_1",
        "timestamp": "1532402927647951",
        "ego_pose": {
            "position": {"x": -1180.8903791765097, "y": 411.3039349319818, "z": 0},
            "heading": {
                "qx": -0.0016977771610471074,
                "qy": 0.011798001930183783,
                "qz": -0.8201446642457809,
                "qw": 0.5720320396729045,
            },
        },
        "default_z": -1,
    },
    {
        "pcd": {"url": pc_urls[1], "type": "pcd"},
        "name": "sample_point_cloud_2",
        "timestamp": "1532402928147847",
        "ego_pose": {
            "position": {"x": -1176.6769733781416, "y": 409.7431520488096, "z": 0},
            "heading": {
                "qx": -0.0013288675751650467,
                "qy": 0.013917029415654661,
                "qz": -0.818440972734323,
                "qw": 0.5744205119534879,
            },
        },
        "default_z": -1,
    },
]

attributes = {"frames": frames}

sample = client.add_sample(dataset, name, attributes)

Set the point cloud types

Are you sure you want to delete this label set? All labels in the label set will be deleted as well.

Are you sure you want to delete 0 samples? The corresponding labels will be deleted as well.