Superannotate Databricks Connector

SuperAnnotate is the cornerstone of your data labeling pipeline. It brings you a cutting-edge annotation tool for all types of data including image, video, text, LiDAR, audio, and more.

This Python package provides a set of utilities for working with SuperAnnotate data on Databricks. It includes functionality to process SuperAnnotate data and save it to Delta tables.

Features

Convert superannotate annotation data to Apache Spark™ Data Frames. Project types supported: - Vector - Text

Example notebooks.

Copy the notebooks in the demo folder to your databricks workspace to get started with SuperAnnotate quickly!

Installation

pip install superannotate_databricks_connector

Tests

Run tests by building the Dockerfile.test file using

docker build -f Dockerfile.test -t test_package .

If you are running the tests for the first you first have to build the base dockerfile containing pyspark.

docker build -f Dockerfile.spark -t spark_docker_base .

Build package

In the main directory, run the following to generate a .whl file.

python -m build

Usage

First import the required function

from superannotate_databricks_conector.vector import get_vector_dataframe
from superannotate import SAClient

You can then convert your annotations to a spark dataframe

sa = SAClient(token="<TOKEN>")
annotations = sa.get_annotations("<PROJECT_NAME>)
df = get_vector_dataframe(annotations, spark)