This repository provides an open-source framework for free-form image inpainting, implementing two published deep learning methods: DeepFill v1 with contextual attention (CVPR 2018) and DeepFill v2 with gated convolution (ICCV 2019 Oral). The system uses generative adversarial networks to fill missing regions in images with realistic content. It is built on TensorFlow and includes pretrained models along with training code.