With the exponential increase in the number of images and videos that are shared online and the prevalence of closed-circuit television systems, it is vital to prevent unwanted disclosure of personal identity. As an extension of Lee et al.’s work on differentially pri- vate obfuscation of facial images, this honours project attempts to obfuscate human pos- es. In this context, the privacy guarantee and generative neural networks are studied. Ex- periments are performed on a pose generating framework (PG2) by adding controlled noises to feature maps at various stages of encoding. Due to the Covid-19 outbreak, the experiments were not finished. As a supplement, a survey was done to describe related generative neural networks.