Backend Setup

The MetaDent App is entirely frontend-based and does not include a backend server. It requires LFSS (Lite File Storage Service) for managing file uploads and storage.

The LFSS is essentially a lightweight file storage server that provides RESTful APIs for uploading, downloading, and managing files. It act like a simplified version of AWS S3 or similar services.

LFSS Server Setup

To set up LFSS for use with the MetaDent App, follow these steps on the server where you want to host (your local machine for this guide):

Install LFSS:

pip install 'lfss[all]'

Add your first user:

lfss-user add <username> --admin --permission private --max-storage 1T

This will result in a hashed access token being generated. Make sure to save it somewhere safe.

Run the server:

lfss-serve

By default, the server will run on port 8000. You can change this by providing the --port option.

The server should now be running and accessible at http://localhost:8000.

You can visit http://localhost:8000/.panel/ in your browser to access the LFSS panel, where you can manage users, view uploaded files, and monitor storage usage.

Our server setup is complete! Simple 😃

Prepare Files for Labeling

The MetaDent App requires a specific directory structure for the images to be labeled. You need to create a directory for your project and upload your images there. The structure should look like this (example):

metadent-project/
├─ images/
│  ├─ image1.jpg
│  ├─ image2.png
│  └─ ...
└─ metadata/
   ├─ image1/
   │  ├─ info.json
   │  └─ (Labeled files will be stored here)
   ├─ image2/
   │  ├─ info.json
   │  └─ (Labeled files will be stored here)
   └─ ...

The info files should be JSON files containing metadata about each image. Essential fields include:

• file_name: The name of the image file.
• height: The height of the image in pixels.
• width: The width of the image in pixels.
• source: The original source of the image (e.g., dataset name, website, etc.).

You can first create this structure locally and then upload it to the LFSS server via the web panel.

Now we've completed the backend setup for the MetaDent App! You can now configure the app to connect to your LFSS server and start labeling images. 🙌

Extra Details

File and Directory Naming Conventions

• The name for the main project directory, image directory, and metadata directory can be customized as needed. Need to be specified when configuring the MetaDent App.
• The name for each image file can be anything, as long as it is unique within the image directory. But please be sure the name of each metadata subdirectory (e.g., image1, image2) should match the corresponding image filename without the extension.

Upload the Files via Command Line (Alternative to Web Panel)

You can use the LFSS CLI tool to upload your project directory to the server:

# first setup LFSS CLI with your server info
export LFSS_ENDPOINT=http://localhost:8000
export LFSS_TOKEN=<your-generated-token>

# then upload your project directory, using -j for parallel uploads
lfss up local/project/path/ <username>/<your-project>/ -j 4

Auxillary AI Service (Optional)

The MetaDent App also supports integration with auxillary AI services for generating initial labels or suggestions.

For details on how to develop a custom auxillary AI service and integrate it with the MetaDent App, please refer to the Auxillary AI Service Guide.