Researchers At SenseTime Develop HuMMan: A Multi-Modal 4D Human Dataset For Versatile Sensing And Modeling

Sensing and modeling humans have long been a challenge for the computer vision and computer graphics communities, which provide the foundation for a wide range of applications, including animation, gaming, augmented reality, and virtual reality.

With the introduction of large-scale datasets and the advent of deep learning, tremendous progress in human-centric sensing and modeling has been made. HuMMan, a comprehensive human dataset comprising of 1000 human participants collected in a total of 400k sequences and 60M frames, was recently presented …

computervision dataset human modeling researchers sensetime sensing