What is the Label Studio ML backend?

The Label Studio ML backend is an SDK that lets you wrap your machine learning code and turn it into a web server. The web server can be connected to a running Label Studio instance to automate labeling tasks.

If you just need to load static pre-annotated data into Label Studio, running an ML backend might be overkill for you. Instead, you can import pre-annotated data.

Quickstart

To start using the models, use docker-compose to run the ML backend server.

Use the following command to start serving the ML backend at http://localhost:9090:

git clone https://github.com/HumanSignal/label-studio-ml-backend.git
cd label-studio-ml-backend/label_studio_ml/examples/{MODEL_NAME}
docker-compose up

Replace {MODEL_NAME} with the name of the model you want to use (see below).

Models

The following models are supported in the repository. Some of them work without any additional setup, and some of them require additional parameters to be set.

Check the Required parameters column to see if you need to set any additional parameters.

• Pre-annotation column indicates if the model can be used for pre-annotation in Label Studio:
  you can see pre-annotated data when opening the labeling page or after running predictions for a batch of data.
• Interactive mode column indicates if the model can be used for interactive labeling in Label Studio: see interactive predictions when performing actions on labeling page.
• Training column indicates if the model can be used for training in Label Studio: update the model state based the submitted annotations.

MODEL_NAME	Description	Pre-annotation	Interactive mode	Training	Required parameters
segment_anything_model	Image segmentation by Meta	❌	✅	❌	None
llm_interactive	Prompt engineering with OpenAI, Azure LLMs.	✅	✅	✅	OPENAI_API_KEY
grounding_dino	Object detection with prompts. Details	❌	✅	❌	None
tesseract	Interactive OCR. Details	❌	✅	❌	None
easyocr	Automated OCR. EasyOCR	✅	❌	❌	None
spacy	NER by SpaCy	✅	❌	❌	None
flair	NER by flair	✅	❌	❌	None
bert_classifier	Text classification with Huggingface	✅	❌	✅	None
huggingface_llm	LLM inference with Hugging Face	✅	❌	❌	None
huggingface_ner	NER by Hugging Face	✅	❌	✅	None
nemo_asr	Speech ASR by NVIDIA NeMo	✅	❌	❌	None
mmdetection	Object Detection with OpenMMLab	✅	❌	❌	None
sklearn_text_classifier	Text classification with scikit-learn	✅	❌	✅	None
interactive_substring_matching	Simple keywords search	❌	✅	❌	None
langchain_search_agent	RAG pipeline with Google Search and Langchain	✅	✅	✅	OPENAI_API_KEY, GOOGLE_CSE_ID, GOOGLE_API_KEY

(Advanced usage) Develop your model

To start developing your own ML backend, follow the instructions below.

1. Installation

Download and install label-studio-ml from the repository:

git clone https://github.com/HumanSignal/label-studio-ml-backend.git
cd label-studio-ml-backend/
pip install -e .

2. Create empty ML backend:

label-studio-ml create my_ml_backend

You can go to the my_ml_backend directory and modify the code to implement your own inference logic.

The directory structure should look like this:

my_ml_backend/
├── Dockerfile
├── docker-compose.yml
├── model.py
├── _wsgi.py
├── README.md
└── requirements.txt

Dockefile and docker-compose.yml are used to run the ML backend with Docker. model.py is the main file where you can implement your own training and inference logic. _wsgi.py is a helper file that is used to run the ML backend with Docker (you don't need to modify it). README.md is a readme file with instructions on how to run the ML backend. requirements.txt is a file with Python dependencies.

3. Implement prediction logic

In your model directory, locate the model.py file (for example, my_ml_backend/model.py).

The model.py file contains a class declaration inherited from LabelStudioMLBase. This class provides wrappers for the API methods that are used by Label Studio to communicate with the ML backend. You can override the methods to implement your own logic:

def predict(self, tasks, context, **kwargs):
    """Make predictions for the tasks."""
    return predictions

The predict method is used to make predictions for the tasks. It uses the following:

• tasks: Label Studio tasks in JSON format
• context: Label Studio context in JSON format - for interactive labeling scenario
• predictions: Predictions array in JSON format

Once you implement the predict method, you can see predictions from the connected ML backend in Label Studio.

4. Implement training logic (optional)

You can also implement the fit method to train your model. The fit method is typically used to train the model on the labeled data, although it can be used for any arbitrary operations that require data persistence (for example, storing labeled data in a database, saving model weights, keeping LLM prompts history, etc).

By default, the fit method is called at any data action in Label Studio, like creating a new task or updating annotations. You can modify this behavior from the project settings under Webhooks.

To implement the fit method, you need to override the fit method in your model.py file:

def fit(self, event, data, **kwargs):
    """Train the model on the labeled data."""
    old_model = self.get('old_model')
    # write your logic to update the model
    self.set('new_model', new_model)

with

• event: event type can be 'ANNOTATION_CREATED', 'ANNOTATION_UPDATED', etc.
• data the payload received from the event (check more on Webhook event reference)

Additionally, there are two helper methods that you can use to store and retrieve data from the ML backend:

• self.set(key, value) - store data in the ML backend
• self.get(key) - retrieve data from the ML backend

Both methods can be used elsewhere in the ML backend code, for example, in the predict method to get the new model weights.

Other methods and parameters

Other methods and parameters are available within the LabelStudioMLBase class:

• self.label_config - returns the Label Studio labeling config as XML string.
• self.parsed_label_config - returns the Label Studio labeling config as JSON.
• self.model_version - returns the current model version.
• self.get_local_path(url, task_id) - this helper function is used to download and cache an url that is typically stored in task['data'], and to return the local path to it. The URL can be: LS uploaded file, LS Local Storage, LS Cloud Storage or any other http(s) URL.

Run without Docker

To run without Docker (for example, for debugging purposes), you can use the following command:

label-studio-ml start my_ml_backend

Test your ML backend

Modify the my_ml_backend/test_api.py to ensure that your ML backend works as expected.

Modify the port

To modify the port, use the -p parameter:

label-studio-ml start my_ml_backend -p 9091

Deploy your ML backend to GCP

Before you start:

1. Install gcloud.
2. Initialize billing for your account if it's not activated.
3. Initialize gcloud, enter the following commands and login with your browser:

gcloud auth login

4. Activate your Cloud Build API.
5. Find your GCP project ID.
6. (Optional) Add GCP_REGION with your default region to your ENV variables.

To start deployment:

1. Create your own ML backend
2. Start deployment to GCP:

label-studio-ml deploy gcp {ml-backend-local-dir} \
--from={model-python-script} \
--gcp-project-id {gcp-project-id} \
--label-studio-host {https://app.heartex.com} \
--label-studio-api-key {YOUR-LABEL-STUDIO-API-KEY}

3. After Label Studio deploys the model, you can find the model endpoint in the console.

Troubleshooting

Troubleshooting Docker Build on Windows

If you encounter an error similar to the following when running docker-compose up --build on Windows:

exec /app/start.sh : No such file or directory
exited with code 1

This issue is likely caused by Windows' handling of line endings in text files, which can affect scripts like start.sh. To resolve this issue, follow the steps below:

Step 1: Adjust Git Configuration

Before cloning the repository, ensure your Git is configured to not automatically convert line endings to Windows-style (CRLF) when checking out files. This can be achieved by setting core.autocrlf to false. Open Git Bash or your preferred terminal and execute the following command:

git config --global core.autocrlf false

Step 2: Clone the Repository Again

If you have already cloned the repository before adjusting your Git configuration, you'll need to clone it again to ensure that the line endings are preserved correctly:

1. Delete the existing local repository. Ensure you have backed up any changes or work in progress.
2. Clone the repository again. Use the standard Git clone command to clone the repository to your local machine.

Step 3: Build and Run the Docker Containers

Navigate to the appropriate directory within your cloned repository that contains the Dockerfile and docker-compose.yml. Then, proceed with the Docker commands:

1. Build the Docker containers: Run docker-compose build to build the Docker containers based on the configuration specified in docker-compose.yml.

2. Start the Docker containers: Once the build process is complete, start the containers using docker-compose up.

Additional Notes

• This solution specifically addresses issues encountered on Windows due to the automatic conversion of line endings. If you're using another operating system, this solution may not apply.
• Remember to check your project's .gitattributes file, if it exists, as it can also influence how Git handles line endings in your files.

By following these steps, you should be able to resolve issues related to Docker not recognizing the start.sh script on Windows due to line ending conversions.

Troubleshooting Pip Cache Reset in Docker Images

Sometimes, you want to reset the pip cache to ensure that the latest versions of the dependencies are installed. For example, Label Studio ML Backend library is used as label-studio-ml @ git+https://github.com/HumanSignal/label-studio-ml-backend.git in requirements.txt. Let's assume that it is updated, and you want to jump on the latest version in your docker image with the ML model.

You can rebuild a docker image from scratch with the following command:

docker compose build --no-cache

Troubleshooting Bad Gateway and Service Unavailable errors

You might see these errors if you send multiple concurrent requests.

Note that the provided ML backend examples are offered in development mode, and do not support production-level inference serving.