# Smart, Deep Copy-Paste

## TL;DR

This paper introduces a deep learning system for smart copy-paste that seamlessly blends source and target images by resolving shading and geometric inconsistencies, trained with a novel dataset-agnostic procedure.

## Contribution

It presents a new end-to-end deep convolutional neural network framework for high-quality image copy-paste that generalizes across datasets without requiring labels.

## Key findings

- Outperforms state-of-the-art on face image datasets
- Demonstrates effective high-resolution results on Cityscapes
- Generalizes well to higher resolutions than training data

## Abstract

In this work, we propose a novel system for smart copy-paste, enabling the synthesis of high-quality results given a masked source image content and a target image context as input. Our system naturally resolves both shading and geometric inconsistencies between source and target image, resulting in a merged result image that features the content from the pasted source image, seamlessly pasted into the target context. Our framework is based on a novel training image transformation procedure that allows to train a deep convolutional neural network end-to-end to automatically learn a representation that is suitable for copy-pasting. Our training procedure works with any image dataset without additional information such as labels, and we demonstrate the effectiveness of our system on two popular datasets, high-resolution face images and the more complex Cityscapes dataset. Our technique outperforms the current state of the art on face images, and we show promising results on the Cityscapes dataset, demonstrating that our system generalizes to much higher resolution than the training data.