We present a method for using deep neural networks to interactively select objects in images. Selecting objects in images can be a time consuming task due to the complexities of objects, similarity in object and background colors, difficulty of textures, and so on. We train a neural network to learn this task. Our model responds to the user inputs (i.e. it can select different objects in the same image based on the user input). We show that we can compute highly accurate selections given minimal user input.