DiMoSR: Feature Modulation via Multi-Branch Dilated Convolutions for Efficient Image Super-Resolution

2025-05-27

AI Summary

Lightweight single-image super-resolution (SISR) faces the fundamental trade-off between high reconstruction fidelity and low computational cost. To address this, we propose an explicit feature modulation paradigm based on multi-branch dilated convolutions—designed to replace or complement conventional attention mechanisms—thereby significantly expanding the effective receptive field without increasing model parameters or computational overhead. Our key innovation lies in the first application of learnable multi-branch dilated convolutions for explicit feature modulation, integrated within a lightweight residual architecture and trained end-to-end under pixel-wise supervision. Extensive experiments on standard benchmarks—including Set5, Set14, and Urban100—demonstrate that our method consistently achieves superior PSNR and SSIM scores over state-of-the-art lightweight SISR models, while maintaining or reducing computational complexity. The source code and pre-trained models are publicly available.

Technology Category

• Computer Vision: Multi-modal Vision
• Machine Learning: Multimodal Learning
• Intelligent Robots: Multimodal Perception & Sensor Fusion

Application Category

• Search and Retrieval-Augmented AI: Retrieval-Augmented Generation (RAG) and multi-modal RAG
• Web Mining and Content Analysis: Large pretrained models with web data
• Responsible Web: Data and user privacy-enhancing technologies for the Web

Abstract

Balancing reconstruction quality versus model efficiency remains a critical challenge in lightweight single image super-resolution (SISR). Despite the prevalence of attention mechanisms in recent state-of-the-art SISR approaches that primarily emphasize or suppress feature maps, alternative architectural paradigms warrant further exploration. This paper introduces DiMoSR (Dilated Modulation Super-Resolution), a novel architecture that enhances feature representation through modulation to complement attention in lightweight SISR networks. The proposed approach leverages multi-branch dilated convolutions to capture rich contextual information over a wider receptive field while maintaining computational efficiency. Experimental results demonstrate that DiMoSR outperforms state-of-the-art lightweight methods across diverse benchmark datasets, achieving superior PSNR and SSIM metrics with comparable or reduced computational complexity. Through comprehensive ablation studies, this work not only validates the effectiveness of DiMoSR but also provides critical insights into the interplay between attention mechanisms and feature modulation to guide future research in efficient network design. The code and model weights to reproduce our results are available at: https://github.com/makinyilmaz/DiMoSR

Problem

• Balancing quality and efficiency in lightweight image super-resolution
• Exploring alternatives to attention mechanisms in SISR networks
• Enhancing feature representation with multi-branch dilated convolutions

Innovation

• Multi-branch dilated convolutions enhance features
• Modulation complements attention in lightweight networks
• Wider receptive field maintains computational efficiency