HyperAIHyperAI

Command Palette

Search for a command to run...

5 months ago

Restoration of Analog Videos Using Swin-UNet

Agnolucci Lorenzo ; Galteri Leonardo ; Bertini Marco ; Del Bimbo Alberto

Restoration of Analog Videos Using Swin-UNet

Abstract

In this paper, we present a system to restore analog videos of historicalarchives. These videos often contain severe visual degradation due to thedeterioration of their tape supports that require costly and slow manualinterventions to recover the original content. The proposed system uses amulti-frame approach and is able to deal with severe tape mistracking, whichresults in completely scrambled frames. Tests on real-world videos from a majorhistorical video archive show the effectiveness of our demo system. The codeand the pre-trained model are publicly available athttps://github.com/miccunifi/analog-video-restoration.

Code Repositories

miccunifi/analog-video-restoration
Official
pytorch
Mentioned in GitHub

Benchmarks

BenchmarkMethodologyMetrics
analog-video-restoration-on-tapeAgnolucci et al.
LPIPS: 0.060
PSNR: 34.96
SSIM: 0.940
VMAF: 77.83

Build AI with AI

From idea to launch — accelerate your AI development with free AI co-coding, out-of-the-box environment and best price of GPUs.

AI Co-coding
Ready-to-use GPUs
Best Pricing
Get Started

Hyper Newsletters

Subscribe to our latest updates
We will deliver the latest updates of the week to your inbox at nine o'clock every Monday morning
Powered by MailChimp
Restoration of Analog Videos Using Swin-UNet | Papers | HyperAI