~/talk

VibroSense: Recognizing home activities by deep learning subtle vibrations on an interior ...

on YouTube 06m
Talkat UbiComp / ISWC 2020 · 2020-09-12..09-17· [machine-learning]

speakers

name:Wei Sun
links:none on file
name:Tuochao Chen
links:none on file
name:Jiayi Zheng
links:none on file
name:Zhenyu Lei
links:none on file
name:Lucy Wang
links:none on file
name:Benjamin Steeper
links:none on file
name:Peng He
links:none on file
name:Matthew Dressa
links:none on file
name:Feng Tian
links:none on file
name:Cheng Zhang
links:none on file

resources

video:https://youtu.be/SAntPSGF2r0
slides:none on file

summary

A system uses a single-point Laser Doppler vibrometer on a wall or ceiling to capture subtle structural vibrations, applies Wavelet denoising, STFT into 2‑second windows and feeds 2D spectra to a CNN on 2D spectrograms to classify 18 home activities. Evaluation on 487 samples across five houses (25 hours) yields 96.6% same‑day accuracy, 89.4% after re‑setup, 90.99% simulated online, 92.2% across angle/distance shifts, and 97.5% for toilet stage classification.

AI-written summary · about

chapters (4)

  1. 0:00Introduction
  2. 0:22Why is it important
  3. 3:51Evaluation
  4. 4:57Results

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.