This project aims to construct a photo-realistic version of hand-drawn sketches of human faces using Deep Convolutional Neural Networks, using the CelebA dataset varying extensively in terms of pose, occlusion, illumination and background clutter. - TensorFlow library used to train CNN and OpenCV used for preprocessing and filtering data.