Agentic AI System for Image Restoration | Blur, Low Light, Underwater & Haze Fix

This project shows an intelligent agentic system designed for complex image restoration tasks. It focuses on restoring real-world images suffering from multiple degradations such as motion blur, defocus blur, low light conditions, dehazing, and underwater distortions. It employs a learning-from-exploration approach where the system generates, evaluates, and distills experience to effectively restore images. It supports fine-tuning, data preparation with synthetic degradations, exploration-based learning, and inference pipelines for image restoration. The system aims to advance low-level vision restoration by integrating agentic workflow designs to handle complex degradation scenarios efficiently.