BISTU Computer Science School Has Two Papers Accepted by CVPR 2024

By Thu, 29 Feb 2024 GMT


     Beijing, 29th February 2024 (BISTU)--- Two papers by Song Wenfeng and Hou Xia, members of Intelligent Text and Knowledge Engineering Innovation Team of Computer Science School, were accepted by CVPR 2024, the premier annual computer vision event. The two papers, as achievements of the joint research with State Key Laboratory of Virtual Reality Technology and Systems based at Beihang University, symbolized the outstanding achievements of the Team has made in virtual reality and artificial intelligence.  

  “HOIAnimator: Generating Text-prompt Human-object Animations using Novel Perceptive Diffusion Models” proposes a perceptual diffusion model that can directly generate human-object interaction animations based on textual descriptions, ensuring smooth and consistent animations through a perceptual message passing mechanism; the proposed interactive contact field model provides an in-depth understanding of the nature of human-object interactions, thus creating dynamic and context-aware animations.  

  

   Fig. 1: HOIAnimator: Generating Text-prompt Human-object Animations using Novel Perceptive Diffusion Models 

  “Arbitrary Motion Style Transfer with Multi-condition Motion Latent Diffusion Model” proposes a multi-conditional potential diffusion model applicable to arbitrary movement style migration. The model achieves a harmonious integration of movement essence and style by delicately deconstructing and integrating movement trajectory, content and style. Moreover, the model prioritises the content channel as the core, and smoothly integrates the multi-conditions through the dynamic support of the trajectory and style channels, which enhances the expressiveness and keeps the content intact.  

  

   Fig. 2: Arbitrary Motion Style Transfer with Multi-condition Motion Latent Diffusion Model 

  Intelligent Text and Knowledge Engineering Innovation Team is dedicated to research and application in the field of computer vision, natural language processing, and virtual reality. In-depth cooperation has been conducted with leading research teams such as State Key Laboratory of Virtual Reality Technology and Systems based at Beihang University, Institute of Computing Technology of Chinese Academy of Sciences, Stony Brook University, etc. Members of the Team have chaired a number of state-level or provincial-level projects and published over 30 papers on leading journals.  

  The publication of high-quality papers fully demonstrates the strength and development potential of the Intelligent Text and Knowledge Engineering Innovation Team in the field of artificial intelligence. The team said that in the future it will continue to work together with top academic institutions at home and abroad,unswervingly promote the cutting-edge exploration of virtual reality and artificial intelligence technology, and draw a new blueprint for the combination of virtual reality and artificial intelligence.