CLIP Lens Dataset

Project CLIP Lens contributes with three datasets of synthetic images corresponding to the three use cases proposed in the project. The first use case corresponds to clothing items (a t-shirt, a sweatshirt and a top). In total, there are 57 images. The second use case corresponds to different professions (chef, doctor, firefighter, journalist, police and teacher) in various situations. In total, there are 234 images. The last use case corresponds to everyday objects, (a book, an iron, a mouse, a chair, a mug and a teapot). In total there are 155 images. Each image is saved with its description, to facilitate future training. Additionally, the folder name indicates the version of Stable Diffusion (1, 1.5 or 2.1) with which the images have been generated, e.g., “sweatshirt-1” would correspond to version 1.