Solara Imagery Labeling Interface

Purpose

Label geospatial imagery collaboratively with a lightweight, customizable platform for multiple simultaneous users.

Running the Interface

The intended way to run this interface is using docker compose with the compose.yml. After cloning this repository, navigate to the repository and run the command: docker compose up -d To build the docker image and run the interface.

Settings for running the interface can be modified by editing the file solara-labeler/src/settings.yml.

In order to run this interface with the original MA Orthophoto dataset it was designed for, clone the data generation repository at github.com/ClarkCGA/solara-labeler-datagen. Modify the compose.yml file in the data generation repository to specify your data directory and run the dataset generation script. Then, modify the compose.yml file in this repository to mount the same data directory to the public directory within the container - for example:

    volumes:
      - /path/to/workspace/solara-labeler/:/home/jovyan/solara-labeler
      - /path/to/workspace/data/:/home/jovyan/solara-labeler/src/public/

Architecture

Core Components

1. Frontend Interface

   • Solara for reactive UI components
   • Leafmap for interactive mapping capabilities
   • Progress tracking and navigation controls

2. Data Management

   • LocalTileServer for serving local imagery
   • CSV tracking system for label boundaries and completion status
   • GeoJSON export functionality for annotated regions

3. Application Flow

   • Load imagery from local files using LocalTileServer or from pre-rendered tiles
   • Display imagery in Leafmap with AOI boundary boxes
   • Capture user-drawn ROIs and save to GeoJSON
   • Update tracking CSV to mark completed labels
   • Pre-load next images for seamless transition

Directory Structure

/
├── Dockerfile
├── LICENSE
├── README.md
├── compose.yml
└── src
    ├── pages
    │   ├── 00_home.py
    │   └── 01_interface.py
    ├── public
    └── settings.yml

Data Structures

Labels CSV Format

id,boundary,status,timestamp,username
1,data/imagery/chip1.tif,"[x1,y1,x2,y2]",pending,null
2,data/imagery/chip2.tif,"[x1,y1,x2,y2]",complete,2025-05-20T14:30:00Z

GeoJSON Annotation Format

{
  "type": "FeatureCollection",
  "features": [
    {
      "type": "Feature",
      "properties": {
        "id": 1,
        "label_class": "building",
        "timestamp": "2025-05-20T14:30:00Z"
      },
      "geometry": {
        "type": "Polygon",
        "coordinates": [[[x1, y1], [x2, y2], ...]]
      }
    }
  ]
}