D
Data
Text
Deepfake Generation and Detection Methods in Deep Learning
A survey of the deep-learning methods used to synthesize and to detect manipulated facial imagery and video. On the generation side it covers Generative Adversarial Networks, Variational Autoencoders, few-shot learning, Transformer networks and diffusion models; on the detection side it covers artifact-based, frequency-based, attention and Transformer classifiers and multimodal fusion, together with standardized task definitions, benchmark datasets and evaluation metrics.