Image Editing Assistant

A modular Python application that uses Google's Gemini API to perform intelligent image editing operations including analysis, global adjustments, and local object manipulation.

Features

• 🤖 Intelligent Routing: Automatically determines the right editing approach
• 📊 Image Analysis: Detailed image information and content analysis
• 🌈 Global Edits: Brightness, contrast, saturation, and color temperature adjustments
• 🎯 Local Edits: Object detection and inpainting for targeted modifications
• 💬 Chat-based Editing: Interactive conversational editing sessions

Quick Start

1. Installation

# Clone or download the project files
# Install dependencies using UV (Python package manager)
uv sync

2. Configuration

# Copy the environment template
cp .env.example .env

# Edit .env and add your API key
# .env file:
GEMINI_API_KEY="your_actual_api_key_here"
# Gemini API information: https://gemini.readthedocs.io/en/latest/

Running Scripts

This project uses uv for Python package management. When running Python scripts or managing dependencies, use uv run commands:

# Run the CLI interface (without GUI)
uv run main.py

Gradio UI Interface

Usage

Launch the modern Gradio web interface:

# Launch the web UI
uv run launch_ui.py [--use-gemini-local-edit]

Then open your browser to http://localhost:7860 to access the interface.

Web UI Features:

• 📤 Image Upload: Upload images via drag-and-drop, file browser, or webcam
• 🖌️ Mask Drawing: Built-in brush and eraser tools for precise inpainting masks
• 💬 Chat Interface: Real-time conversation with the AI assistant
• 🔄 Live Updates: See edits applied in real-time
• 📥 Download: Save your edited images instantly
• 🎛️ Image Tools: Crop, flip, rotate, and transform images

Agent Architecture

Agent Responsibilities

Router Agent

• Analyzes user prompts to determine the appropriate editing action
• Routes requests to the appropriate specialized agent: info, global_edit, local_edit, or advanced
• Uses Gemini API to understand user intent and select the best agent for the task

Info Agent

• Analyzes images to provide detailed information about content and characteristics
• Generates concise descriptions (under 150 words) in a single paragraph format
• Identifies objects, scenes, colors, lighting conditions, and other visual elements

Global Edit Agent

• Performs whole-image adjustments like brightness, contrast, saturation
• Handles color temperature, sharpness, and other global parameters
• Applies filters and overall image enhancements

Local Edit Agent

• Performs targeted edits on specific objects or regions
• Handles object detection, selection, and manipulation
• Supports inpainting for object removal and replacement