Human behavior has been studied in detail mainly for action classification and prediction, but there have been few works exploring the generation of novel actions. Generating novel sequences of human motion to form an action has been a challenging problem. These motions can either simulate full body movements, like gait, or part-specific movement, like playing the guitar or phone call, or involve facial expressions, action units, or even mouth movement when a person speaks or reads a text.
With the advent of powerful generative models such as Generative Adversarial Networks (GAN), novel data generation paradigms have become possible, and these networks have shown to be powerful in many image generation tasks. However, many issues remain to be solved especially when passing from the static to the dynamic case and new research problems emerge. For example, what are the appropriate architectures and loss functions to generate dynamic facial expressions, and actions, in 2D or in 3D? Which are the best objective and subjective methods to evaluate the generative models?
We expect generating synthetic and realistic static and dynamic data of humans can have a big impact in several different contexts. A straightforward outcome that developing such techniques could have is that of generating an abundance and variety of new data that could be otherwise difficult, very expensive, and time-consuming to obtain from reality. Such data can be essential in simulation, virtual and augmented reality, and in training more robust learning tools, to cite a few. For example, we could expect new applications in the game and movie industry, where fully synthetic actors could be used in the near future, without the need for explicit modeling. We expect this workshop could help to make a step in these research directions, also focusing on new evaluation methodologies that could make quantitative rather than qualitative the assessment and comparison of generated data. With respect to this latter aspect, we expect the public release of new benchmarks especially in 3D could help to improve a lot of the research in this field. Finally, there is quite a debate in society and in the scientific community too about ethical implications related to generating synthetic and realistic human data. We also aim to have a discussion on these social and ethical implications at the workshop.
For more information, you can find the call for papers here.
The goal of this workshop is to provide contributions to this emerging field of study. Topics of interest of this workshop include but are not limited to:
2D and 3D static and dynamic content generation
2D and 3D facial expression generation
Talking heads generation
Actions generation
Behavior generation
Evaluation of generative models
New benchmarks
Reduce ethical issues in human data generation
Applications (data augmentation, simulation, VA, medical, robotics)
Workshop organized in conjunction with WACV 2021, Winter Conference on Applications of Computer Vision, Waikoloa, Hawaii, Virtual Event, January 5-9, 2021