Backend

Setup

Recommended to use uv to manage virtual environment and pip packages.

First Installation

Create virtual environment:

uv venv

Install dependencies:

uv pip install -r requirements.txt

PostgreSQL + PostGIS

Note: The following names and passwords are only for development purposes, and will be different in production.

Create a PostgreSQL database with PostGIS enabled, then set:

docker run --name utd-postgis \
    -e POSTGRES_USER=utd \
    -e POSTGRES_PASSWORD=utdpass \
    -e POSTGRES_DB=utd_data \
    -p 5432:5432 \
    -d postgis/postgis:16-3.4

If this not the first time running, use now:

docker start utd-postgis

Make sure your environment has the variable DATABASE_URL which corresponds to the new DB.

export DATABASE_URL='postgresql+psycopg://utd:utdpass@localhost:5432/utd_data'

Preprocessing Pipeline

Run both parsing and DB loading in one command:

python3 util/preprocess-data.py ../test_images_labels_targets/test --output data-example

Running the API

If parsed images are not under data-example, point the API to the parser output root:

export PARSED_DATA_DIR='data-example'

uvicorn app.main:app --reload