MobileIE: An Extremely Lightweight and Effective ConvNet for Real-Time Image Enhancement on Mobile Devices

2025-07-02

AI Summary

To address the challenge of real-time image enhancement under resource-constrained mobile scenarios, this paper proposes an ultra-lightweight convolutional network framework. Methodologically, it integrates structural reparameterization, incremental weight optimization, a feature self-transformation module, and a hierarchical dual-path attention mechanism, coupled with a local variance-weighted loss function to enhance perceptual quality. The key contribution lies in achieving unprecedented real-time inference—up to 1100 FPS—with merely ~4K parameters, establishing a new state-of-the-art speed–accuracy trade-off across mainstream image enhancement tasks. The model is exceptionally compact and computationally efficient, enabling low-latency on-device deployment. Experimental results demonstrate that our approach maintains competitive image quality while significantly outperforming existing lightweight methods in both efficiency and effectiveness.

Technology Category

• Computer Vision: Learning & Optimization for CV
• Machine Learning: Learning on the Edge & Model Compression
• Search and Optimization: Learning to Search

Application Category

• Search and Retrieval-Augmented AI: Efficiency and scalability of Web search engines
• Systems and Infrastructure for Web, Mobile and WoT: Applied ML and AI for Web-based mobile applications
• User Modeling, Personalization and Recommendation: On-Device user modeling, personalization, and recommendation

Abstract

Recent advancements in deep neural networks have driven significant progress in image enhancement (IE). However, deploying deep learning models on resource-constrained platforms, such as mobile devices, remains challenging due to high computation and memory demands. To address these challenges and facilitate real-time IE on mobile, we introduce an extremely lightweight Convolutional Neural Network (CNN) framework with around 4K parameters. Our approach integrates reparameterization with an Incremental Weight Optimization strategy to ensure efficiency. Additionally, we enhance performance with a Feature Self-Transform module and a Hierarchical Dual-Path Attention mechanism, optimized with a Local Variance-Weighted loss. With this efficient framework, we are the first to achieve real-time IE inference at up to 1,100 frames per second (FPS) while delivering competitive image quality, achieving the best trade-off between speed and performance across multiple IE tasks. The code will be available at https://github.com/AVC2-UESTC/MobileIE.git.

Problem

• Deploying deep learning models on mobile devices efficiently
• Achieving real-time image enhancement with minimal parameters
• Balancing speed and performance in image enhancement tasks

Innovation

• Lightweight CNN with 4K parameters
• Reparameterization and Incremental Weight Optimization
• Feature Self-Transform and Dual-Path Attention