Video enhancement system leveraging ESRGAN and OpenCV to upscale frames and revitalize videos with superior visual quality.
Existing ESRGAN implementations for upscaling video files are based on PyTorch. There exists a couple of them which uses Tensorflow with FFMPEG directly.
This repository is built on Tensorflow and OpenCV. Useful when you'd want to integrate this on the top of a project built already using the above mentioned tech without the need for additional dependencies.
Side by side comparison of an original 144p video upscaled 4x using BICUBIC interpolation on left and ESRGAN on right, achieving a PSNR value of 37.10
You can either fine-tune your own on a base model, or use one from TensorFlow hub.
Use conda / pip to create a Python 3.7 virtual environment.
pip install -r requirements.txt
Note for Windows users: Use numpy==1.19.3