watermark-removal: a machine learning based image inpainting tool for removing watermarks

What it solves

It removes watermarks from images, aiming to produce results that are indistinguishable from the original, watermark-free version of the image.

How it works

The project uses a machine learning-based image inpainting methodology. It is inspired by research on Contextual Attention and Gated Convolution to fill in the missing pixels where the watermark was located.

Who it’s for

Users who need to clean up images by removing unwanted watermarks using deep learning techniques.

Highlights

• Uses machine learning for high-quality image inpainting.
• Inspired by CVPR 2018 and ICCV 2019 research papers.
• Provides Docker support for easier deployment.
• Supports specific watermark types, such as istock.