An Efficient Segmentation Based Pipeline For Image Inpainting

This repository contains the code for an integrated pipeline used to inpaint images. The details of the approach can be found in report.pdf.

Requirements

• Install python3
• Install Tensorflow
• Install tensorflow toolkit neuralgym (run pip install git+https://github.com/JiahuiYu/neuralgym)
• Install skimage
• Install OpenCV
• Download the pretrained models for foreground from (https://drive.google.com/drive/folders/1mGomKU3YYtx-b3G27VAHekUmMCANXFx_?usp=sharing) and place it under fg/model_logs
• Download the pretrained tagger from (https://drive.google.com/open?id=15LRt2MPTUUBUFCIakLthSYNZ01Ie3yke) and place it under tag/models

Usage

usage : run_algo.py input_image output_image [image_mask]

positional arguments:
  input image           Path to image to be infilled
  output_image          Path to ouput fill
  image_mask            Path to the mask

Authors

• Abhinav Gupta
• Vaishnav S. Menon
• Suyash Shukla