KidusMa/amharic-prompt-image-enhancer

+A sophisticated photo editing application that leverages Large Language Models (LLMs) to enhance images using Amharic prompts. This application combines multiple AI technologies to provide intelligent photo editing capabilities. + +## Features + +### 🤖 AI-Powered Technologies +- **NLLB Translation**: High-accuracy Amharic-to-English translation using Hugging Face's NLLB model +- **LLM Agent Planning**: Task decomposition using LLaMA-3 inspired architecture for breaking down editing requests +- **Generative Editing**: Advanced image manipulation using Stable Diffusion InstructPix2Pix +- **Multimodal Evaluation**: Image-prompt evaluation using vision-language models for quality assessment +- **Basic Image Processing**: OpenCV/Pillow integration for fundamental adjustments + +### 📱 User Interface +- **Flutter Frontend**: Modern, responsive mobile and web interface +- **Amharic Support**: Native Amharic text input with proper font rendering +- **Real-time Preview**: Side-by-side comparison of original and edited images +- **Manual Controls**: Sliders for brightness, contrast, and saturation adjustments +- **Full-screen Viewing**: Zoom and pan capabilities for detailed image inspection + +### 🔧 Backend Capabilities +- **RESTful API**: Flask-based backend with comprehensive endpoints +- **Image Upload**: Support for various image formats with automatic preprocessing +- **Batch Processing**: Efficient handling of multiple editing operations +- **Error Handling**: Robust error management with user-friendly feedback + +## Architecture + +``` +┌─────────────────┐ ┌──────────────────┐ ┌─────────────────┐ +│ Flutter App │───▶│ Flask Backend │───▶│ AI Models │ +│ │ │ │ │ │ +│ • Image Upload │ │ • Translation │ │ • NLLB │ +│ • Amharic Input │ │ • Task Planning │ │ • InstructPix2Pix│ +│ • Results View │ │ • Image Process │ │ • LLaMA-3 │ +│ • Manual Edit │ │ • Eva …