testing-cv

Image Quality Assessment (IQA)

• NIMA
• TRIQ
• Explainable
• Auto correction
  • Exposure
  • White-balance
  • Denoising
• Detection
  • ColorCast

Flicker Detection

• Preprocessing
  • Optical Flow
  • Image Normalization
  • VGG19 feature extractor
• Modeling
  • End to End VGG BiDirectional LSTM
  • End to End VGG 3DViT

Flicker Outlier Detection Results

Install

Ubuntu

Tested on

1. Docker Ubuntu Official Image 18.04
2. Google Cloud Platform - Ubuntu 16.04, 18.04, 20.04

Anaconda

conda on Linux

# create env from yml file
conda env create -f executable.yml
# activate env
conda activate "enviorment name here"

Docker

• Prerequisite: Docker

# Clone this repo
git clone https://github.com/hc07180011/testing-cv.git
cd testing-cv/flicker_detection/flicker_detection

# Build the container
docker build -t flicker_detection_runner .

# Run the container with a specific input data (put it under data/ directory)
docker run --rm -e data=data/test_data.mp4 -v $PWD/data:/app/data -it flicker_detection_runner

Latest Updates

• 2022/11/9 Flicker Report
• 2022-2023 Google CV Research Proposal