RT-Focuser: A Real-Time Lightweight Model for Edge-side Image Deblurring

2025-12-26

AI Summary

Motion blur severely degrades image quality, hindering real-time edge applications such as autonomous driving and UAV perception. To address this, we propose a lightweight U-shaped deblurring network tailored for resource-constrained edge devices. Our method introduces three key innovations: (1) a Lightweight Deblurring (LD) block to reduce computational redundancy; (2) a Multi-Level Integrated Aggregation (MLIA) module to enhance multi-scale feature representation; and (3) a Cross-source Fusion (X-Fuse) block enabling efficient encoder–decoder information interaction. The resulting model achieves a compact footprint of only 5.85 million parameters and 15.76 GMACs, with single-frame inference latency of 6 ms (<7 ms) on both GPU and mobile platforms—exceeding 140 FPS. It attains a PSNR of 30.67 dB, striking a strong balance between accuracy and speed. The source code is publicly available.

Technology Category

• Machine Learning: Learning on the Edge & Model Compression
• Computer Vision: Multi-modal Vision
• Intelligent Robots: Multimodal Perception & Sensor Fusion

Application Category

• Systems and Infrastructure for Web, Mobile and WoT: Applied ML and AI for Web-based mobile applications
• Responsible Web: Machine-in-the-loop, human agency and autonomy
• User Modeling, Personalization and Recommendation: On-Device user modeling, personalization, and recommendation

Abstract

Motion blur caused by camera or object movement severely degrades image quality and poses challenges for real-time applications such as autonomous driving, UAV perception, and medical imaging. In this paper, a lightweight U-shaped network tailored for real-time deblurring is presented and named RT-Focuser. To balance speed and accuracy, we design three key components: Lightweight Deblurring Block (LD) for edge-aware feature extraction, Multi-Level Integrated Aggregation module (MLIA) for encoder integration, and Cross-source Fusion Block (X-Fuse) for progressive decoder refinement. Trained on a single blurred input, RT-Focuser achieves 30.67 dB PSNR with only 5.85M parameters and 15.76 GMACs. It runs 6ms per frame on GPU and mobile, exceeds 140 FPS on both, showing strong potential for deployment on the edge. The official code and usage are available on: https://github.com/ReaganWu/RT-Focuser.

Problem

• Develops a lightweight model for real-time image deblurring on edge devices
• Addresses motion blur challenges in autonomous driving and UAV perception
• Balances speed and accuracy for deployment on mobile and GPU platforms

Innovation

• Lightweight U-shaped network for real-time deblurring
• Edge-aware feature extraction with Lightweight Deblurring Block
• Progressive decoder refinement via Cross-source Fusion Block