56 datasets found

Conditional Adversarial Generative Flow for Controllable Image Synthesis

The LLVIP dataset contains registered images in two domains of visible light and infrared light.

A dataset of 121 video sequences of common household objects, captured by GoPro Hero9 action cameras at 30 FPS and 1920 × 1080 resolution.

The dataset used in the Discriminator Rejection Sampling (DRS) algorithm, consisting of a mixture of 25 2D-Gaussian distributions.

The dataset used in the paper is LSUN bedroom and church-outdoor datasets (64×64).

The dataset used in the paper is not explicitly described, but it is mentioned that the authors used the Stable Diffusion Model (SDM) v1.5.

The dataset used to evaluate the capacity of a denoising diffusion probabilistic model to reproduce spatial context in medical imaging.

The dataset used to evaluate the capacity of a denoising diffusion probabilistic model to reproduce spatial context in medical imaging.

The CUB dataset is a large collection of images of birds, each image is a 299x299 RGB image, and there are 11,778 training images and 5,994 testing images.

The SinGAN dataset is used for one-shot image synthesis.

The FastGAN dataset is used for low-shot image synthesis.

A high-quality and diverse dataset for evaluating the effectiveness of fake image detection methods, including 6 GAN and 6 diffusion models, to generalize unknown forgery...

The UPNA Synthetic dataset consists of 12 videos for each of 10 subjects; 120 videos in total with 38,800 frames.

The YaleFace dataset consists of gray-scale frontal face images of 38 subjects under 64 illumination conditions.

The dataset used in this paper for generative model evaluation, consisting of 4.1M images from 41 state-of-the-art generative models spanning diffusion models, GANs, variational...

The dataset used in the paper is the LSUN bedrooms dataset, a large-scale image dataset.

A dataset of images of buildings and landmarks, used for training and testing image-based rendering and view synthesis algorithms.

The DDPM dataset consists of data from 86 subjects attempting to answer questions deceptively. Interviews were recorded at 90 frames-per-second for more than 10 minutes on average.

A dataset for person image generation. The approach uses a combination of 2D and 3D techniques to generate person images.

A high-quality synthetic dataset created for evaluating generative models