Google Vision Tags

A script to get AI-generated tags from Google image recognition API

This script takes an input csv containing artwork data and outputs a csv with labels from the Google Vision API.

As of January 2023, this code is experimental and currently in development.

Requirements

• Python 3.6+
• macOS

Python 3.6+ requires the installation of an SSL certificate, to install in terminal run:

/Applications/Python\ 3.11/Install\ Certificates.command

or to skip installation and use a deprecated ssl context set --use-deprecated-ssl-context to true.

Usage

Use --help flag to see a full list of arguments and explanations:

python3 ./aic_google_vision_script.py --help

The only required arguments are api-key and input-csv. You can generate and manage Google Cloud API keys in the Google Cloud web console. Here is a full list of arguments with the values used to generate the output delivered with completion of this work:

python3 ./aic_google_vision_script.py \
    --api-key API_KEY \
    --input-csv input/All_Paintings_CSV_2022DEC23.csv \
    --output-csv output/All_Paintings_CSV_2022DEC23_output.csv \
    --failed-row-csv output/All_Paintings_CSV_2022DEC23_failed_rows.csv \
    --log-level info \
    --max-labels 50 \
    --batch-size 4 \
    --label-filters "Art,Visual Arts,Artist,Paint" \
    --use-deprecated-ssl-context false \
    --throttle-time 5

Configuration

Following is a description of all of the configurations that can be entered when using the script:

api-key

Type: String

Required

A key provided by Google to authorize access to their API.

Example:

--api-key XXXX

input-csv

Type: String

Required

Relative path to input CSV Keep all input files in the /input folder. CSV should include the following columns:

• Title
• Artist Title
• Id
• Website URL
• IIIF Image URL

Example:

--input-csv input/artworks_20230101.csv

output-csv

Type: String

Default: output# where # is an incremented number

Relative path to output CSV. Keep all output files in the /output folder.

Example:

--output-csv output/artworks_20230101_output.csv

failed-row-csv

Type: String

Default: failed_rows# where # is an incremented number

All rows that failed from input csv are copied here for easier retry. Keep all output files in the /output folder.

Example:

--failed-row-csv output/artworks_20230101_failed_rows.csv

log-level

Type: String

Default: info

Sets verbosity of execution logging. Allowed values are [debug, info, warn, error].

Example:

--log-level info

max-labels

Type: Integer

Default: 50

The maximum number of labels for each input image.

Example:

--max-labels 50

batch-size

Type: Integer

Default: 6

The number of images included in each request to Google vision API.

Example:

--batch-size 4

label-filters

Type: String

Default:

Comma separated labels to filter out for all images. (eg art, painting, etc).

Example:

--label-filters "Art,Visual Arts,Artist,Paint"

use-deprecated-ssl-context

Type: Boolean

Default: false

Use a deprecated ssl context.

Example:

--use-deprecated-ssl-context true

throttle-time

Type: Float

Default: 2.5

Amount of time in seconds to wait between processing each batch, needed to avoid usage limits on API.

Example:

--throttle-time 5

starting-row-number

Type: Integer

Default: 1

Row number of input-csv to begin processing on.

Example:

--starting-row-number 4

ending-row-number

Type: Integer

Default: System's maximium integer size

Row number of input-csv to end processing on.

Example:

--ending-row-number 50

overwrite-output-files

Type: Boolean

Default: false

Whether to overwrite output files, if false output will be appended to files if they exist.

Example:

--overwrite-output-files true

Contributing

We encourage your contributions. Please fork this repository and make your changes in a separate branch. To better understand how we organize our code, please review our version control guidelines.

# Clone the repo to your computer
git clone [email protected]:your-github-account/google-vision-tags.git

# Enter the folder that was created by the clone
cd google-vision-tags

# Start a feature branch
git checkout -b feature/good-short-description

# ... make some changes, commit your code

# Push your branch to GitHub
git push origin feature/good-short-description

Then on github.com, create a Pull Request to merge your changes into our main branch.

This project is released with a Contributor Code of Conduct. By participating in this project you agree to abide by its terms.

We welcome bug reports and questions under GitHub's Issues. For other concerns, you can reach our engineering team at [email protected]

Acknowledgements

Initial development by Elliot Korte.

Licensing

This project is licensed under the GNU Affero General Public License Version 3.