scaneo 2024.6.4.post2

A labelling tool for satellite imagery

Create your own training datasets from satellite imagery, effortlessly and faster than ever.

SCANEO is an AI-powered web tool for smart labeling of satellite data training datasets. It can generate labels on its own, and its model can also be retrained with user-validated labels, creating a feedback loop that increases accuracy and speeds up labeling. In this way, SCANEO benefits both models and data sets simultaneously, and promotes the proliferation of artificial intelligence models applied to Earth observation data.

Why SCANEO?

The shortage of suitable and accessible training datasets used to train AI models applied to Earth observation data is just one of the many problems faced by users who want to apply artificial intelligence to satellite data. Furthermore, the acquisition and labeling of EO data is complicated and expensive, slowing the advancement of AI in EO and limiting its potential compared to other fields.

This is how SCANEO emerges, a smart labeling web application for training sets with satellite data, powered through artificial intelligence and active learning.

Installation

SCANEO is simply a Python package that can be installed using pip.

pip install scaneo

Is is recommended to upgrade the package regularly, in order to get the latest changes.

pip install scaneo --upgrade

The library requires Python >= 3.8

Usage

SCANEO allows to launch the labelling web application through CLI commands. The first thing you can do is run the help command, which will give you a list of all the available commands in the CLI.

scaneo --help

You can launch scaneo with a single command, just specifying the path to the folder where the sallite data is stored using the --data flag:

scaneo --data <path-to-you-folder>

This will launch the UI, which will be accessible on your localhost:8000.

Documentation and tutorials

To view the documentation, launch the UI and go to localhost:8000/docs. There you will find all the detailed SCANEO documentation, with advanced examples of usage, videos and tutorials.

Build

This repository contains the source code for SCANEO.

• scaneo: includes the CLI, the library and the API.
• ui: includes the web application UI.

The CLI runs the API, which in turns serves the static files for the UI.

Development

Run the API with the CLI:

cd scaneo
python main.py run --reload --data <folder>

Then, run the UI:

cd ui
yarn dev

Remember to create the .env file from .env.example cp .env.example .env

Production

In the UI, create .env.production with an empty API_URL.

Build the UI, copy the build inside scaneo and build the python package:

make build v=<version>
make publish

It is needed to install mkdocs with pip install mkdocs-material

Notes

Do not add scaneo/ui to gitignore since the build process will fail (missing entry folder).