Abstract: Estimating 3D hand meshes from RGB images robustly is a highly desirable task, made challenging due to the numerous degrees of freedom, and issues such as self similarity and occlusions. Previous methods generally either use parametric 3D hand models or follow a model-free approach. While the former can be considered more robust, e.g. to occlusions, they are less expressive. We propose a hybrid approach, utilizing a deep neural network and differential rendering based optimization to demonstrably achieve the best of both worlds. In addition, we explore Virtual Reality (VR) as an application. Most VR headsets are nowadays equipped with multiple cameras, which we can leverage by extending our method to the egocentric stereo domain. This extension proves to be more resilient to the above mentioned issues. Finally, as a use-case, we show that the improved image-model alignment can be used to acquire the user’s hand texture, which leads to a more realistic virtual hand representation.
Realistic.Hands_compressed.mp4
arxiv: https://arxiv.org/abs/2108.13995?context=cs
coming soon
The paper describes a newer and improved approach, however the original thesis includes background information and more detailed descriptions that may be of interest.