Home

Awesome

<img src="frontend/images/pixano_logo.png" alt="Pixano" height="100"/>

License Live Demo License

What is PIXANO ?

Pixano is a web-based smart-annotation tool for computer vision applications. The modules are driven by artificial intelligence, which assists the human user with annotation tasks and accelerate the annotation process. Try some of our features online!

pixano.gif

Pixano App

Table of Contents

1. Installation

Using Docker Image

The easiest way to get up-and-running is to install Docker. Then, you should be able to download and run the pre-built image using the docker command line tool. Find out more about the pixano image on its Docker Hub page.

Here's the simplest way you can run the Pixano application using docker:

sudo docker run -it --rm -v "$PWD":/data -p 3000:3000 pixano/pixano-app

The path where you run this command must contain the data you want to annotate.

NB: This path is defined as your workspace.

Optional: create an alias

In practice, we suggest you setup an alias called pixano to automatically expose the folder containing your specified image, so the script can read it and store results where you can access them. This is how you can do it in your terminal console on OSX or Linux:

# Setup the alias. Put this in your .bashrc file so it's available at startup.
# Note that the --network host works only on Linux, use explicit port mapping for Windows and Mac
alias pixano='function ne() { if [ -d "$(pwd)/$1" ]; then DATA="$(pwd)/$1" && shift; else DATA="$(pwd)"; fi; sudo docker run --init -it --rm --network host -v "$DATA":/data pixano/pixano-app $@; }; ne'

# Now run pixano using alias with workspace as argument
pixano ./data-test --port 3001
# or omit workspace and use current directory by default
# pixano

Install from source (for developers)

Install global dependencies

# Make sure you have curl installed
sudo apt install curl
# Then download and execute the Node.js 10.x installer
curl -sL https://deb.nodesource.com/setup_12.x | sudo -E bash -
# Once the installer is done doing its thing, you will need to install (or upgrade) Node.js
sudo apt install nodejs
# Make sure the version is now correct
nodejs --version
npm install -g npm@6.10.0

You can read this nice [introduction](https://codeburst.io/the-only-nodejs-introduction-youll-ever-need-d969a47ef219) to NodeJS in case you're curious on how it works.

ATTENTION: node version 16 is not compatible for now

Install application dependencies

npm run deps
Using a local pixano-element

If you want to use custom pixano-element modules from local path instead of the NPM registry, link them as explained below:

# Install application dependencies and local pixano-elements
npm run installLocalElements --path=$PIXANO_ELEMENTS_PATH

NB: Make sure you have the git repository of pixano-elements next to the pixano-app folder and that you have followed the pixano-elements build instructions before running the above commands.

If this command breaks your local pixano-elements demo, this command will repear it:

cd $PIXANO_ELEMENTS_PATH
npm run bootstrap

Build the application

# Bundle the application using Webpack
# This will create a build folder containing all the sources to be served
npm run build

Run the application

In the command prompt, type in pixano /path/to/your/workspace from the root folder and hit enter. (or alternatively : node cli/pixano /path/to/your/workspace)

NB: Make sure when typing this command that the workspace (/path/to/your/workspace) contains all of the data you want to use.

Type in pixano --help to get the options available in your current version of Pixano.

2. Usage

After running Pixano-App, you’ll see something similar to this:

   ┌────────────────────────────────────────────────────────────────────────┐
   │                                                                        │
   │   Serving   /path/to/your/workspace                                    │
   │                                                                        │
   │   - Local:            http://localhost:3000                            │
   │   - On Your Network:  http://xxx.xxx.x.xx:3000                         │
   │                                                                        │
   └────────────────────────────────────────────────────────────────────────┘

Open your browser and hit http://localhost:3000. You should see the login page of the application.

pixano-elements

First authentication is: username: admin password: admin.

Configure your first annotation project

Before annotating, configure your project by following our admin's guide. You will be able to:

Start annotating

Once a task is defined, you (or your annotators) will be able to annotate your dataset. See our annotator's guide for your first steps.
Our plugins' guide will help you in the use of your current task's specific plugin.

Export your annotations

Get your annotations and use them for any external application easily:

Standalone vs distributed usage

Pixano-app can be used standalone on a single machine. In this case, the "admin" can also directly annotate and validate his datasets. See our admin's guide for more details.

Pixano-app is also developed to enable a distributed work:

3. Advanced usage

Import predictions

If you want to analyze predictions from your last detector or use these predictions as a pre-annotation, you can import these predictions as existing annotations by using our annotation format.

Import/Export annotation format

data-test   
│
│───images
│   │    xxx.jpg
│   └─── yyy.jpg
│       
└───annotations
    │─── task1.json
    └─── task1
        │    xxx.json
        └─── yyy.json

The task1.json file contains global task settings (task type, task categories, image folder, etc) and its corresponding task1 folder contains an annotation file for each image. To prepare those files check the import documentation.

Build docker from sources

To create a docker image of the application, you can use the standard docker command:

# You can change `pixano` by your choosen image name
sudo docker build -t pixano/pixano-app:my-tag .

If you used a local pixano-element, build the application (step 1.b) and then run:

# You can use the local Dockerfile if the build folder already exists
sudo docker build -t pixano/pixano-app:my-tag -f Dockerfile-local .

4. Contributing

If you tested Pixano and identified some issues or think some useful features are missing, please open an issue.

If you want to edit the application to your liking, fork this repository!

If you want to contribute more actively to the project, feel free to write your patches or new features and make a pull request!

Pixano architecture: Pixano-app and Pixano-elements

pixano.gif

Pixano-app is a monorepo built on top of web components dedicated to annotation (developed in a separate repo: pixano-elements):

Pixano-elements provides a wide set of smart and re-usable web components to build highly customizable image and video annotation tools: 2D and 3D bounding boxes, polygons, segmentation masks, customizable labels, label temporal propagation, etc. Pixano-app relies on these web components.

Some documentation to get started