Annotate Lab is an open-source application designed for image annotation, comprising two main components: the client and the server. The client, a React application, is responsible for the user interface where users perform annotations. On the other hand, the server, a Flask application, manages persisting the annotated changes and generating masked and annotated images, along with configuration settings. More information can be found in our documentation.
- Project Structure
- Dependencies
- Setup and Installation
- Running the Application
- Running Tests
- Code Formatting
- Usage
- Settings
- Configurations (Optional)
- Demo
- Auto Bounding Box Selection with Segment Anything Model (SAM)
- Outputs
- YOLO Format
- Troubleshooting
- Contributing
- License
- Reporting Security Issues
- Acknowledgment
Project Structure [documentation page]
annotation-lab/
βββ client/
β βββ public/
β βββ src/
β βββ package.json
β βββ package-lock.json
β βββ ... (other React app files)
βββ server/
β βββ db/
β βββ tests/
β βββ venv/
β βββ app.py
β βββ requirements.txt
β βββ ... (other Flask app files)
βββ README.md
- public/: Static files and the root HTML file.
- src/: React components and other frontend code.
- package.json: Contains client dependencies and scripts.
- db/: Database-related files and handlers.
- venv/: Python virtual environment (not included in version control).
- tests/: Contains test files.
- app.py: Main Flask application file.
- requirements.txt: Contains server dependencies.
Dependencies [documentation page]
Setup and Installation [documentation page]
- Navigate to the
client
directory:cd client
- Install the dependencies:
npm install
- Navigate to the
server
directory:cd server
- Create and activate a virtual environment:
python3 -m venv venv source venv/bin/activate # On Windows use `venv\Scripts\activate`
- Install the dependencies:
pip install -r requirements.txt
- Navigate to the
client
directory:cd client
- Install the dependencies:
npm start
The application should now be running on http://localhost:5173.
- Navigate to the
server
directory:cd server
- Activate the virtual environment:
source venv/bin/activate # On Windows use `venv\Scripts\activate`
- Start the Flask application:
flask run
The server should now be running on http://localhost:5000.
Navigate to the root directory and run the following command to start the application:
docker-compose build
docker-compose up -d #running in detached mode
The application should be running on http://localhost.
Running Tests [documentation page]
The client tests are located in the client/src
directory and utilize .test.js
extensions. They are built using Jest and React Testing Library.
cd client
npm install
npm test
This command launches the test runner in interactive watch mode. It runs all test files and provides feedback on test results.
The server tests are located in the server/tests
directory and are implemented using unittest.
cd ../server
python -m venv venv
source venv/bin/activate # On Windows use `venv\Scripts\activate`
pip install -r requirements.txt
python3 -m unittest discover -s tests -p 'test_*.py'
This command discovers and runs all test files (test_*.py
) in the server/tests
directory using unittest.
Code Formatting [documentation page]
- Code Formatter: Prettier
- Configuration File:
.prettierrc
- Command: Run
npm run format
oryarn format
to format client-side code using Prettier.
- Code Formatter: Black
- Configuration File:
pyproject.toml
- Command: Run
black .
to format server-side code using Black.
- Open your web browser and navigate to http://localhost:5173.
- Use the user interface to upload and annotate images.
- The annotations and other interactions will be handled by the Flask server running at http://localhost:5000.
Settings [documentation page]
One can configure the tools, tags, upload images and do many more from the settings.
Configurations (Optional) [documentation page]
You can customize various aspects of Annotate-Lab through configuration settings. To do this, modify the config.py
file in the server
directory or the config.js
file in the client
directory.
# config.py
MASK_BACKGROUND_COLOR = (0, 0, 0) # Black background for masks
SAM_MODEL_ENABLED = False # Segment Anything Model for auto bounding box selection
// config.js
const config = {
SERVER_URL, // url of server
UPLOAD_LIMIT: 500, // image upload limit
OUTLINE_THICKNESS_CONFIG : { // outline thickness of tools
POLYGON: 2,
CIRCLE: 2,
BOUNDING_BOX: 2
},
SAM_MODEL_ENABLED: false, // displays button that allows auto bounding box selection
SHOW_CLASS_DISTRIBUTION: true // displays annotated class distribution bar chart
};
Demo V2.0
Auto Bounding Box Selection with Segment Anything Model (SAM)[documentation page]
Selection of bounding box automatically is made possible with the Segment Anything Model (SAM). One can toggle this feature from the configuration of server and client. When enabled, a wand icon will appear in the toolbar. Clicking the wand icon will initiate auto-annotation and display the results
Outputs [documentation page]
Sample of annotated image along with its mask and settings is show below.
{
"orange.png": {
"configuration": [
{
"image-name": "orange.png",
"regions": [
{
"region-id": "13371375927088525",
"image-src": "http://127.0.0.1:5000/uploads/orange.png",
"class": "Print",
"comment": "",
"tags": "",
"points": [
[
0.5863691595741748,
0.7210152721281337
],
[
0.6782101128815677,
0.6587584627896123
],
[
0.7155520389516067,
0.5731553499491453
],
[
0.7286721751383771,
0.40065210740699225
],
[
0.7518847237765094,
0.352662483541882
],
[
0.6862840428426572,
0.2307428985872776
],
[
0.6045355019866261,
0.1581099543590026
],
[
0.533888614827093,
0.13476365085705708
],
[
0.44204766151970004,
0.13476365085705708
],
[
0.3441512607414899,
0.17886222413850975
],
[
0.2957076809749529,
0.23852499975459276
],
[
0.2523103074340969,
0.3163460114277445
],
[
0.2129498988737856,
0.418810343464061
],
[
0.20891293389324087,
0.5121955574718431
],
[
0.22506079381541985,
0.6016897208959676
],
[
0.2563472724146416,
0.6652435470957082
],
[
0.30378161093604245,
0.7197182552669145
],
[
0.3683730506247584,
0.7819750646054359
],
[
0.4057149766947973,
0.8066183849686005
],
[
0.46223248642242376,
0.776786997160559
],
[
0.5308608910916844,
0.7586287611034903
]
]
}
],
"color-map": {
"Apple": [
244,
67,
54
],
"Orange": [
33,
150,
243
]
}
}
]
}
}
YOLO Format [documentation page]
YOLO format is also supported by A.Lab. Below is an example of annotated ripe and unripe tomatoes. The entire dataset can be found on Kaggle. In this example, 0
represents ripe tomatoes and 1
represents unripe ones.
The label of the above image are as follows:
0 0.213673 0.474717 0.310212 0.498856
0 0.554777 0.540507 0.306350 0.433638
1 0.378432 0.681239 0.223970 0.268879
Applying the generated labels we get following results.
Normalization process of YOLO annotations [documentation page]
To convert non-normalized bounding box coordinates (xmax, ymax, xmin, ymin) to YOLO format (xcenter, ycenter, width, height):
Image Credit: Leandro de Oliveira
# Assuming row contains your bounding box coordinates
row = {'xmax': 400, 'xmin': 200, 'ymax': 300, 'ymin': 100}
class_id = 0 # Example class id (replace with actual class id)
# Image dimensions
WIDTH = 640 # annotated image width
HEIGHT = 640 # annotated image height
# Calculate width and height of the bounding box
width = row['xmax'] - row['xmin']
height = row['ymax'] - row['ymin']
# Calculate the center of the bounding box
x_center = row['xmin'] + (width / 2)
y_center = row['ymin'] + (height / 2)
# Normalize the coordinates
normalized_x_center = x_center / WIDTH
normalized_y_center = y_center / HEIGHT
normalized_width = width / WIDTH
normalized_height = height / HEIGHT
# Create the annotation string in YOLO format
content = f"{class_id} {normalized_x_center} {normalized_y_center} {normalized_width} {normalized_height}"
print(content)
The above conversion will give us YOLO format string.
0 0.46875 0.3125 0.3125 0.3125
Troubleshooting [documentation page]
- Ensure that both the client and server are running.
- Check the browser console and terminal for any errors and troubleshoot accordingly.
- Verify that dependencies are correctly installed.
If you would like to contribute to this project, please fork the repository and submit a pull request. For major changes, open an issue first to discuss your proposed changes. Additionally, please adhere to the code of conduct. More information about contributing can be found here.
This project is licensed under the MIT License.
If you find a security vulnerability in annotate-lab, please read our Security Policy for instructions on how to report it securely.
This project is detached from idapgroup's react-image-annotate, which is licensed under the MIT license, and it uses some work from image_annotator.