~/talk

VisMimic: Integrating Motion Chain in Feedback Video Generation for Motor Coaching

on YouTube 13m
Talkat UIST 2025 · 2025-09-28..10-01· [ai-assisted-development, design-collaboration]

speakers

name:Liqi Cheng
links:none on file
name:Xiao Xie
links:none on file
name:Yiwei Peng
links:none on file
name:Minghao Feng
links:none on file
name:Yuchen He
links:none on file
name:Anqi Cao
links:none on file
name:Yihong Wu
links:none on file
name:Hui Zhang
links:none on file
name:Yingcai Wu
links:none on file

resources

video:https://youtu.be/U46vgE9WckQ
slides:none on file

summary

VisMimic builds an authoring pipeline that automatically extracts key poses by detecting extrema in joint trajectories, converts motions into a motion chain with angular/positional/temporal kinematic constraints, and synthesizes animated feedback using PoseFix, a diffusion-based avatar generator, and fallback full-body IK. A 3D fitness dataset powers constraint-weighting across a parent–child joint hierarchy and greedy decoding to rank feedback candidates. User studies report positive ratings, SUS in the acceptable range, and no significant difference versus expert videos, while reconstruction accuracy and object interactions remain limitations.

AI-written summary · about

chapters (6)

  1. 0:00Introduction and motivation
  2. 2:32Literature review and design
  3. 6:05System interface and demo
  4. 8:54User studies and results
  5. 9:57Limitations and future work
  6. 10:19Q&A session

related talks

~/search
type to search 6,249 conferences, 138,987 speakers and 185,172 talks
~/help
cmdctrl k
search every conference, speaker and series
/
focus the filter search
h
home
c
conferences
s
speakers
e
series
t
light / dark — the status bar button does the same
?
this help
esc
close this, then clear every filter — or just the search, while you are typing in it

Ignored while you are typing. Pages without a filter rail have no search to focus.