<![CDATA[This study presents a Systematic Literature Review (SLR) on the development of attention-based deep learning methods for Single Image Super-Resolution (SISR) for the period 2022–2026. Using the PRISMA 2020 framework, 53 articles were selected from 1,116 identified articles from Scopus, IEEE Xplore, ScienceDirect, and SpringerLink. The study results show that hybrid CNN-Transformer architectures with attention mechanisms dominate the current trend, with channel attention being the most widely used (71.7%). Application domains include medical imaging, remote sensing, surveillance, multimedia, and industry. Key challenges identified include model complexity, limited generalization, and the trade-off between reconstruction quality and inference efficiency. This study provides a comprehensive roadmap for developing more adaptive and efficient attention-based SISR systems.Penelitian ini menyajikan Systematic Literature Review (SLR) terhadap perkembangan metode deep learning berbasis attention mechanism untuk Single Image Super-Resolution (SISR) pada periode 2022–2026. Menggunakan kerangka PRISMA 2020, sebanyak 53 artikel dipilih dari 1.116 artikel teridentifikasi dari Scopus, IEEE Xplore, ScienceDirect, dan SpringerLink. Hasil kajian menunjukkan bahwa arsitektur hibrida CNN-Transformer dengan attention mechanism mendominasi tren terkini, dengan channel attention sebagai jenis yang paling banyak digunakan (71,7%). Domain aplikasi mencakup pencitraan medis, pengindraan jauh, surveillance, multimedia, dan industri. Tantangan utama yang teridentifikasi meliputi kompleksitas model, keterbatasan generalisasi, dan trade-off antara kualitas rekonstruksi dan efisiensi inferensi. Penelitian ini memberikan peta jalan komprehensif bagi pengembangan sistem SISR berbasis attention yang lebih adaptif dan efisien.]]>
Copyrights © 2026
