Utils for Label Studio

The cvtk package provides a command-line utility for generating Label Studio ML backend source code so trained detection or segmentation models can assist annotation.

To enable these utilities to communicate with Label Studio, several environment variables must be set. We recommend setting the following variables before using any of these utilities.

export LABEL_STUDIO_LOCAL_FILES_SERVING_ENABLED=true
export LABEL_STUDIO_LOCAL_FILES_DOCUMENT_ROOT="/Users/USERNAME/path/to/ws"
export LABEL_STUDIO_BASE_DATA_DIR="/U/path/to/labelstudio-data"
export LABEL_STUDIO_URL="http://localhost:8080"
export LABEL_STUDIO_API_KEY="f6dea26f0a0f81883e04681b4e649c600fe50fc"

LABEL_STUDIO_BASE_DATA_DIR specifies the directory where Label Studio stores its data. The path of this directory is differ depending on the system environment and startup options of Label Studio. You can use the following command to find out the path of this directory, start a Python interpreter and execute the commands below.

>>> import os
>>> from label_studio.core.utils.params import get_base_data_dir
>>> os.path.abspath(get_base_data_dir())
~/.local/share/label-studio

LABEL_STUDIO_URL specifies the URL of the Label Studio server. Just copies the URL displayed in the web browser when accessing Label Studio.

LABEL_STUDIO_API_KEY specifies the API key for accessing Label Studio. You can find the API key in Label Studio by navigating to the user profile page (by clicking on the user icon in the top right corner of the screen) and then selecting the “Account” tab. The API key is displayed in the “API Key” section of the page.

Annotation Export

Annotations from Label Studio can be exported from the Label Studio web interface. Export the annotations in COCO format when preparing data for the detection or segmentation tutorials.

ML-Assisted Labeling

The cvtk package allows users to integrate trained machine learning models with Label Studio for ML-Assisted Labeling. This enables models to assist in the annotation process within Label Studio. To do this, use the cvtk deploy-ls-mlbackend command, which generates the necessary source code and supporting files for backend integration.

Assume that a source code file det.py for object detection has already been created using cvtk (as described in the Object Detection Tutorial), and the trained model weights are saved in ./outputs/strawberry.pth.

Then, run the following command, which generates the backend source code and related files in the directory specified by the --project argument.

cvtk deploy-ls-mlbackend \
    --project lsbackend \
    --source det.py \
    --label ./labels.txt \
    --model ./outputs/strawberry.py \
    --weights ./outputs/strawberry.pth

Source files will be generated in the lsbackend directory. To start the backend server, run:

python lsbackend/mlbackend.py --host 0.0.0.0 --port 8080