hitbuyi / point-cloud-annotator

Web annotation tool for point clouds

Point Cloud Annotator (in dev)

A web based annotation tool for 3D computer vision tasks. The tool is developed in the context of cashier-less convenience store research by Suning. It supports point clouds (.pcd) for the annotation task of pose estimation, tracking, and reID. It is a SPA developed with React and three.js.

🎥 VIDEO DEMO: Point Cloud Annotator Demo

For the annotation task of detection and semantic segmentation, please refer to Semantic Segmentation Editor by Hitachi Automotive and Industry Lab.

How to run

Install Node.js

Download a pre-built installer for your platform from https://nodejs.org/en/download/.

Download the repo

git clone https://github.com/zexihan/point-cloud-annotator.git

Start the application

cd point-cloud-annotator
npm install
npm start

Open http://localhost:3000 to view it in the browser.

The page will reload if you make edits.
You will also see any lint errors in the console.

(Optional) Edit the configs.json.

By default, point clouds are served from a subfolder in public/data/pcd. Sample person point clouds are included in public/data/pcd/person.

{
    "set_nm": "person",
    "begin_fid": 0,
    "end_fid": 6,
    "pcd_folder": "./data/pcd",
    "bbox_folder": "./data/bbox",
    "keypoints_folder": "./data/keypoints",
    "keypoints": [
        {"label": "left_ear", "color": "#9A3254"},
        {"label": "right_ear", "color": "#E85D88"},
        {"label": "left_shoulder", "color": "#FF8815"},
        {"label": "right_shoulder", "color": "#FFE832"},
        {"label": "left_elbow", "color": "#1FB93E"},
        {"label": "right_elbow", "color": "#1AAEAA"},
        {"label": "left_wrist", "color": "#83BCFF"},
        {"label": "right_wrist", "color": "#208BFB"},
        {"label": "left_hip", "color": "#54351A"},
        {"label": "right_hip", "color": "#5D36E1"}
    ]
}

How to use

Control

axis helper: X Y Z

left mouse button + move: Pan the map

right mouse button + move: Rotate the view

mouse wheel: Zoom up and down

a/d: Previous/Next frame

+/-: Increase/Decrease point size

c: Change color

f: Mark/Unmark

shift + left mouse button: Add/Remove key point

PCD support

• Supported input PCD format: ASCII
• Supported input fields: x, y, z, rgb (optional integer)