CASIA_Dataset_Creator has all the codes which are required to extract masks for the CASIA dataset. extractmask.py has to be run first and followed by fillmask.py. Please set proper paths. COCO_Dataset_Creator has all codes related to creating masks for coco images. For handling COCO images you have to install their python API. You have to download their annotations and the dataset separately. Change the annotation file path in the cococreation.py. This code will create masks for random objects in images and store them in a folder. I used cocomaskadded.py to add those masks to the images to remove specified regions. Inpainting_codes has codes which I used for Inpainting. These donot inpaint large regions. I also used some deep learning based methods but I donot have the model right now. For feathering I have an adobe dataset which I can share offline. They have their own codes to create the dataset. In patch_creator folder, there are codes for extracting patches from images. Though this should be done in a better way in my experience than what I have written. TF_Codes/Testing has codes for testing. TF_Codes/Training has codes for training. We have to improve on your work: Median Filters CNN [Yashas Paper] How to use K effectively: what about small tampered regions(hence not good for localization) Localization: [Learning Rich Features for Image Manipulation Detection] Using 2 streams of Image one RGB and one noisy. Robust to compression and Resizing What happens on resizing images