Getting Started With Annotation

Trainer currently supports annotating images and videos. First we need to setup the database. Using a database instead of simply storing files helps working with complex, structured data.

First, install postgresql (From conda or from here:). Follow some guide from google to setup an empty database. We recommend to test the connection using psql or pgadmin4 now.

Execute some code that tries to use the database:

python ./tutorials/db.py

In your user directory a new folder .trainer should pop up. Update the connection string db_con.

{
   "big_bin_dir": "C:\\Users\\rapha\\.trainer\\big_bin",
   "db_con": "postgresql+psycopg2://postgres:admin@127.0.0.1:5432/trainer"
}

Now, Import data using the CLI tools. Data might be exported earlier from someone else or imported using several file formats from disk. Here we assume you do not already have a dataset which was exported using trainer.

At this point, start the GUI to start annotating:

trainer annotate -n testset

Generator Usage Workflow

Subject generator -> Preprocessing -> Model optimization