we propose a novel approach for object removals with the techniques of object detection, image inpainting and temporal loss optimization. To implement a deep neural network-based algorithm to detect an object, remove it and fill the gap with a smooth background in a video. The ability of the DNNs and its latest flavor, GANs (Generative Adversarial Networks) to in-paint an image has been extensively studied and showcased, but still there are very less focus on DNN based video inpainting algorithm due to its complexity and temporal dependency between frames. This project tries to implement a novel algorithm to modify an existing image inpainting method and add temporal smoothing constraints to achieve a visually pleasing in-painted video.