PulseExploreJournal ClubResearchersJournals
Instagram
HomeJournal ClubExplore
Synapse
⌘+K
Synapse
February 19, 2026Scientific ReportsOpen Access

Research on super-resolution reconstruction of construction images based on attention mechanism and generative adversarial networks

View Full Paper
Ask AI
Bookmark
Share

Authors

QCQinhuang ChenGHGongyu HouDWDandan Wang

Discussion

Loading...

Member takes

Overview

Proposed super-resolution model improves image quality and inference speed in construction, suggesting new monitoring capabilities.

Key Points

  • The research aims to enhance the quality of construction images using super-resolution techniques.
  • Developed a reconstruction model with attention mechanisms and generative adversarial networks
  • Implemented a Multi-scale Shallow Feature Extraction module
  • Utilized SORRDB for efficient feature mapping and modulation
  • Employed Swin Transformer for the discriminator to capture detailed features
  • Achieved PSNR of 28.063 dB and SSIM of 0.825 on the SODA dataset
  • Obtained LPIPS score of 0.197 indicating improved image realism
  • Reached an inference speed of 32.0 FPS on an NVIDIA GeForce RTX 4070

Cite This Study

Chen et al. (2026) studied this question.

synapsesocial.com/papers/6996712d80e1323b05ec038chttps://doi.org/10.1038/s41598-026-40613-4
View Full Paper
Ask AI
Bookmark
Share