~/talk

Exemplar based experience transfer

on YouTube 22m
Talkat IUI 2019 · 2019-03-17..03-20· [design-collaboration, component-design]

speakers

name:Paridhi Maheshwari
links:none on file
name:Nitish Bansal
links:none on file
name:Surya Dwivedi
links:none on file
name:Rohan Kumar
links:none on file
name:Pranav Manerikar
links:none on file
name:Balaji Vasan Srinivasan
links:none on file

resources

video:https://youtu.be/3O1TFApSTPQ
slides:none on file

summary

System extracts banner templates with a pixel-level fully convolutional neural network segmentation plus a neighborhood search to form rectangular blocks, compared against a combined-detection baseline using Mask R-CNN and found FCNN+postprocessing more precise. User content is placed and tuned by learning weighted layout energies (alignment, spread, overlap) with an inverse optimization and a simulated annealing proposal sampler. Crowdsourced A/B tests on Mechanical Turk show machine results match target layout metrics though designers still score higher; runtime ≈1–1.5 minutes per banner.

AI-written summary · about

chapters (8)

  1. 0:00Intro
  2. 1:20Idea
  3. 2:44Content
  4. 3:32Larger objective
  5. 3:59Our solution
  6. 5:23Sample banner
  7. 7:10Template extraction
  8. 18:02Questions

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.