Image classification app, built with Databricks DB Connect "V2" and Dash 🔥

Prerequisites

• [Local machine] JVM (I'm using JVM 11 in this example)
• [Local machine] Poetry
• [Local machine] Databricks profile configured via databricks-cli
• [Databricks] Databricks All-purpose Cluster with DBR 13.X or higher and UC enabled

How-to

1. Install package and it's dependencies:

poetry install

2. Create a table with image metadata (check the possible arguments in the conf/config.yaml file):

poetry run frontend \
    --config-path="${PWD}/conf" \
    profile=some-databricks-profile \
    cluster_name=some-cluster-name

3. Run the app:

poetry run python \
    db_connect_v2_image_classification/frontend/app.py \
    --config-path="${PWD}/conf" \
    profile=some-databricks-profile \
    cluster_name=some-cluster-name

4. Open http://0.0.0.0:8050 and enjoy your app ✨

Technologies used

• Databricks and DB Connect V2
• hydra for configuration management
• poetry for package management and project setup
• black for code formatting
• ruff for linting
• isort for import analysis
• Dash by Plotly as frontend framework
• Tailwind CSS
• Daisy UI for ready-to-use components