metadata

title: OmniSealBench
emoji: 🔐
colorFrom: blue
colorTo: indigo
sdk: docker
app_port: 5000
pinned: false
license: mit

OmniSealBench

🔐 A comprehensive benchmark for watermarking techniques with a modern web interface.

Overview

OmniSealBench is a benchmarking tool for evaluating various watermarking techniques across different media types including images and audio. This application provides:

Interactive leaderboards for comparing watermarking methods
Visual examples of various watermark attacks and their effects
Easy-to-use interface for exploring benchmark results

Features

Responsive UI: Clean, modern interface that works across devices
Sortable Leaderboard: View and filter benchmark results with advanced sorting
Example Browser: Visualize examples of watermarking techniques and attacks
Multi-Modal Support: Support for both image and audio watermarks

Project Structure

omnisealbench/
├── backend/                 # Flask backend
│   ├── app.py               # Main Flask application
│   ├── api.py               # API endpoints
│   ├── data_processor.py    # Data handling utilities
│   ├── requirements.txt     # Python dependencies
│   ├── Dockerfile.dev       # Development Dockerfile
│   └── start.sh             # Production startup script
├── frontend/                # React frontend
│   ├── src/                 # Source code
│   │   ├── components/      # React components
│   │   ├── api.js           # API client
│   │   ├── App.jsx          # Main application component
│   │   ├── index.jsx        # Entry point
│   │   └── styles.css       # Application styles
│   ├── public/              # Static assets
│   ├── package.json         # Node.js dependencies
│   ├── vite.config.js       # Vite configuration
│   └── Dockerfile.dev       # Development Dockerfile
├── data/                    # Benchmark data files
│   ├── image_benchmark.csv
│   ├── audio_benchmark.csv
│   └── ...
├── examples/                # Example files for visualization
│   ├── image/
│   └── audio/
├── Dockerfile               # Multi-stage build for production
├── README.md                # Project documentation
├── setup_dev.sh             # Development setup script
├── start_dev.sh             # Start development environment
├── stop_dev.sh              # Stop development environment
├── test_build.sh            # Test production build
├── clean_docker.sh          # Clean up Docker resources
└── prepare_deploy.sh        # Prepare for deployment

Prerequisites

Docker (for containerized deployment and development)
Node.js 18+ and npm (for local frontend development)
Python 3.10+ (for local backend development)

Getting Started

Using Docker (Recommended)

Clone the repository:

git clone https://github.com/yourusername/omnisealbench.git
cd omnisealbench

Build and run the application with Docker:
```
./test_build.sh
```
Access the application at http://localhost:5000

Development Environment

Set up the development environment:
```
./setup_dev.sh
```
Start the development environment:
```
./start_dev.sh
```
Access the frontend at http://localhost:3000 and backend at http://localhost:5000
When finished, stop the development environment:
```
./stop_dev.sh
```

Deployment

This application is designed to be deployed using Docker to various platforms.

Quick Deployment

Prepare for deployment:
```
./prepare_deploy.sh
```
Test the build locally:
```
./test_build.sh
```
Follow platform-specific deployment instructions in DEPLOYMENT.md

For detailed deployment instructions for HuggingFace Spaces, AWS, Google Cloud, and Azure, please refer to the Deployment Guide.

API Documentation

The backend provides the following API endpoints:

GET /api/benchmarks - List available benchmarks
GET /api/leaderboard - Get leaderboard data
GET /api/columns - Get available columns for a benchmark
GET /api/examples - Get examples for a specific model
GET /api/attacks - Get available attacks for a benchmark

Converting from Gradio

This project was converted from a Gradio-based HuggingFace space to a Flask/React application. The key changes included:

Replacing Gradio UI components with React components
Creating a Flask API to serve data previously handled by Gradio backend
Implementing a containerized deployment with Docker

Contributing

Contributions are welcome! Please feel free to submit a Pull Request.

License

This project is licensed under the MIT License - see the LICENSE file for details.

Acknowledgements

Flask - Backend framework
React - Frontend library
Vite - Frontend build tool
Pandas - Data processing

Searching

When searching is enabled, a textbox will appear in the top left corner of the leaderboard. Users will be able to display rows that match their search query.

Searching follows the following rules:

Multiple queries can be separated by a semicolon ;.
Any subquery is matched against the primary search column by default.
To match against a secondary search column, the query must be preceded by the column name and a colon (:), e.g. Name: Maria.
The returned rows are those that match against ANY primary search column and ALL secondary search columns.

You can configure searching with the search_columns parameter. It's value can be:

a list: In which case the first element is the primary search column and the remaining are the secondary search columns.
A SearchColumns instance. This lets you specify the primary and secondary columns explicitly as well as customize the search textbox appearance.

Demo

import pandas as pd
import gradio as gr
from gradio_leaderboard import Leaderboard, SearchColumns

with gr.Blocks() as demo:
    Leaderboard(
        value=pd.DataFrame({"name": ["Freddy", "Maria", "Mark"], "country": ["USA", "Mexico", "USA"]}),
        search_columns=SearchColumns(primary_column="name", secondary_columns="country",
                                     placeholder="Search by name or country. To search by country, type 'country:<query>'",
                                     label="Search"),
    )

demo.launch()

Filtering

You can let users filter out rows from the leaderboard with the filter_columns parameter. This will display a series of form elements that users can use to select/deselect which rows are displayed.

This parameter must be a list but it's elements must be:

a string: Corresponding to the column name you'd like to add a filter for
a ColumnFilter: A special class for full control of the filter's type, e.g. checkboxgroup, slider, or dropdown, as well as it's appearance in the UI.

If the type of the ColumnFilter is not specified, a heuristic will be used to choose the most appropriate type. If the data in the column is numeric, a slider will be used. If not, a checkboxgroup will be used.

Demo

import pandas as pd
import gradio as gr
from gradio_leaderboard import Leaderboard, ColumnFilter

with gr.Blocks() as demo:
    Leaderboard(
        value=pd.DataFrame({"name": ["Freddy", "Maria", "Mark"], "country": ["USA", "Mexico", "USA"],
                            "age": [25, 30, 35], "score": [100, 200, 300]}),
        filter_columns=[
            "name",
            ColumnFilter("country", type="dropdown", label="Select Country 🇺🇸🇲🇽"),
            ColumnFilter("age", type="slider", min=20, max=40, greater_than=True),
            ColumnFilter("score", type="slider", min=50, max=350, greater_than=True)],
    )

demo.launch()

`Leaderboard`

Initialization

name	type	default	description
`value`	`pd.DataFrame \| None`	`None`	Default value to display in the DataFrame. If a Styler is provided, it will be used to set the displayed value in the DataFrame (e.g. to set precision of numbers) if the `interactive` is False. If a Callable function is provided, the function will be called whenever the app loads to set the initial value of the component.
`datatype`	`str \| list[str]`	`"str"`	Datatype of values in sheet. Can be provided per column as a list of strings, or for the entire sheet as a single string. Valid datatypes are "str", "number", "bool", "date", and "markdown".
`search_columns`	`list[str] \| SearchColumns`	`None`	See Configuration section of docs for details.
`select_columns`	`list[str] \| SelectColumns`	`None`	See Configuration section of docs for details.
`filter_columns`	`list[str \| ColumnFilter] \| None`	`None`	See Configuration section of docs for details.
`hide_columns`	`list[str] \| None`	`None`	List of columns to hide by default. They will not be displayed in the table but they can still be used for searching, filtering.
`latex_delimiters`	`list[dict[str, str \| bool]] \| None`	`None`	A list of dicts of the form {"left": open delimiter (str), "right": close delimiter (str), "display": whether to display in newline (bool)} that will be used to render LaTeX expressions. If not provided, `latex_delimiters` is set to `[{ "left": "$$", "right": "$$", "display": True }]`, so only expressions enclosed in $$ delimiters will be rendered as LaTeX, and in a new line. Pass in an empty list to disable LaTeX rendering. For more information, see the [KaTeX documentation](https://katex.org/docs/autorender.html). Only applies to columns whose datatype is "markdown".
`label`	`str \| None`	`None`	The label for this component. Appears above the component and is also used as the header if there are a table of examples for this component. If None and used in a `gr.Interface`, the label will be the name of the parameter this component is assigned to.
`show_label`	`bool \| None`	`None`	if True, will display label.
`every`	`float \| None`	`None`	If `value` is a callable, run the function 'every' number of seconds while the client connection is open. Has no effect otherwise. The event can be accessed (e.g. to cancel it) via this component's .load_event attribute.
`height`	`int`	`500`	The maximum height of the dataframe, specified in pixels if a number is passed, or in CSS units if a string is passed. If more rows are created than can fit in the height, a scrollbar will appear.
`scale`	`int \| None`	`None`	relative size compared to adjacent Components. For example if Components A and B are in a Row, and A has scale=2, and B has scale=1, A will be twice as wide as B. Should be an integer. scale applies in Rows, and to top-level Components in Blocks where fill_height=True.
`min_width`	`int`	`160`	minimum pixel width, will wrap if not sufficient screen space to satisfy this value. If a certain scale value results in this Component being narrower than min_width, the min_width parameter will be respected first.
`interactive`	`bool \| None`	`None`	if True, will allow users to edit the dataframe; if False, can only be used to display data. If not provided, this is inferred based on whether the component is used as an input or output.
`visible`	`bool`	`True`	If False, component will be hidden.
`elem_id`	`str \| None`	`None`	An optional string that is assigned as the id of this component in the HTML DOM. Can be used for targeting CSS styles.
`elem_classes`	`list[str] \| str \| None`	`None`	An optional list of strings that are assigned as the classes of this component in the HTML DOM. Can be used for targeting CSS styles.
`render`	`bool`	`True`	If False, component will not render be rendered in the Blocks context. Should be used if the intention is to assign event listeners now but render the component later.
`wrap`	`bool`	`False`	If True, the text in table cells will wrap when appropriate. If False and the `column_width` parameter is not set, the column widths will expand based on the cell contents and the table may need to be horizontally scrolled. If `column_width` is set, then any overflow text will be hidden.
`line_breaks`	`bool`	`True`	If True (default), will enable Github-flavored Markdown line breaks in chatbot messages. If False, single new lines will be ignored. Only applies for columns of type "markdown."
`column_widths`	`list[str \| int] \| None`	`None`	An optional list representing the width of each column. The elements of the list should be in the format "100px" (ints are also accepted and converted to pixel values) or "10%". If not provided, the column widths will be automatically determined based on the content of the cells. Setting this parameter will cause the browser to try to fit the table within the page width.