# VFRAME Check API

The VFRAME Check API Service uses perceptual hash to disambiguate similar images as well as provide an image-based search engine.

## Quick Start

- Install Python 3.x (use Miniconda or your choice)
- Install nginx, MySQL

```
apt install git nginx mysql-server mysql-client
mysql_secure_installation
mysql -u root -p
```

Create MySQL user:

```
CREATE DATABASE vframe_check;
CREATE USER 'vframe_check'@'localhost' IDENTIFIED BY 'some_new_password';
GRANT ALL PRIVILEGES ON vframe_check.* to 'vframe_check'@'localhost';
```

Clone the repo and add a .env file:

```
DB_HOST=localhost
DB_NAME=vframe_check
DB_USER=vframe_check
DB_PASS=some_new_password
```

The initial dataset can be hosted locally for now - make sure files are accessible inside `check/static/`.  These paths will be added directly to the dataset.

```
python cli_phash.py import -i 'static/sample_set_test_01/images/*'
```

Build the Javascript frontend:

```
npm run build
```

Run the server:

```
cd check
python cli_flask.py run
```

The server will be running on http://0.0.0.0:5000/

### Endpoints

#### POST /v1/match

Check if an image is in the database.  If no images are found within the threshold, this image will be added to the database.

Form parameters:

- `threshold` (default: 6) - Minimum similarity threshold.  This is the Hamming distance used for phash comparisons.
- `limit` (default: 1) - Number of results to return.
- `url` - Image URL to fetch and test (will be added if not found!)
- `q` (file) - Uploaded file to test (will not be added)

#### POST /v1/similar

Find similar images to a query image.

Form parameters:

- `threshold` (default: 20) - Minimum similarity threshold.
- `limit` (default: 10) - Number of results to return.
- `url` - Image URL to fetch and test
- `q` (file) - Uploaded file to test