Image Inpainting is a very important topic in the day and age of consuming digital media and design. Inpainting is a process where the damaged or missing parts of artwork is filled to complete it. This report presents the analysis and conclusions of the Image Inpainting Using Partially Convoluted Autoencoder Decoder models and Bilinear GANs. The task is to paint a masked area of an image in accordance to the overall image and generate a fine grained output that improves the accuracy to 2-3%. The goal is to achieve an image that looks realistic to the human eye. The task can be tackled in many ways, the authors of the report chose two methods, the first one being an Autoencoder decoder approach for keras where they used partially convoluted layers for achieving the results. The CIFAR-10 dataset was used to train the model. The model had a dice coefficient of 0.604. The other proposed model is the GAN model. GAN is used because it can generate more refined and visually aesthetic results.