qulishen

qulishen / Flickerformer

Public

The official code for the paper "It Takes Two: A Duet of Periodicity and Directionality for Burst Flicker Removal".

36
0
100% credibility
Found Mar 14, 2026 at 25 stars -- GitGems finds repos before they trend. Get early access to the next one.
Sign Up Free
AI Analysis
Python
AI Summary

Flickerformer removes flicker artifacts from burst photo sequences using a transformer-based image restoration model.

How It Works

1
📱 Snap flickering bursts

You take quick photo bursts on your phone, but notice annoying light flickers ruining the consistency.

2
🔍 Find Flickerformer

You discover this simple photo fixer online, promising smooth, flicker-free results from your bursts.

3
💾 Grab the tool

Download the ready app and sample cleaner to your computer.

4
🖼️ Load your photos

Easily add your flickering burst photos – it recognizes the sequences automatically.

5
▶️ Hit clean

Press go and relax as it processes your bursts, blending them into steady perfection.

✨ Flicker gone forever

Admire your beautifully consistent photos, ready to print or share without distractions.

Sign up to see the full architecture

4 more

Sign Up Free

Star Growth

See how this repo grew from 25 to 36 stars Sign Up Free
Repurpose This Repo

Repurpose is a Pro feature

Generate ready-to-use prompts for X threads, LinkedIn posts, blog posts, YouTube scripts, and more -- with full repo context baked in.

Unlock Repurpose
AI-Generated Review

What is Flickerformer?

Flickerformer cleans flicker artifacts from burst image sequences captured under unstable lighting or rolling-shutter cameras. Feed it three-frame bursts—like from mobile low-light shots—and it outputs flicker-free results by fusing frames with phase correlation and wavelet guidance. Python-based with PyTorch, it includes training scripts, evaluation metrics like PSNR/SSIM/LPIPS, and a BurstDeflicker dataset from Kaggle.

Why is it gaining traction?

It targets burst flicker removal precisely, using periodicity and directionality priors that beat generic restoration methods on visuals and metrics. Ready-to-run dist_train.sh for multi-GPU, test.py for inference, and evaluate.py for benchmarks make experimentation fast. As the official GitHub repository for a CVPR 2026 paper, it draws researchers seeking state-of-the-art baselines.

Who should use this?

Computer vision devs processing phone camera bursts for HDR or night mode apps. Researchers benchmarking image restoration on flicker-heavy datasets. Teams handling real-world video frame sequences needing quick deflickering without ghosting.

Verdict

Worth a spin for burst flicker tasks—solid results on the niche dataset—but 18 stars and 1.0% credibility signal early maturity; docs are paper-focused, tests minimal. Grab from official GitHub releases page and validate on your data.

(187 words)

Sign up to read the full AI review Sign Up Free

Similar repos coming soon.