Have you ever wondered how you would look like if you were a Simpsons character? To answer this question I thought about trying to build a CycleGAN (https://github.com/junyanz/CycleGAN) that would translate human faces to Simpsons faces. To train the model I constructed this datasets, which contains Simpsons faces, extracted from seasons 25-28. The process utilized a simple face recognition algorithm that I have been experimenting with (https://github.com/KostasTok/Simpsons-Face-Detector). Cropped images are as extracted directly from the video files. Simplified images were built as a by-product of the face recognition algorithm. They only present the eyes and skin of the recognized face. All images are 200x200 pixels, each centered on a single face. The images of the two sets corresponded numerically.