Controlled desktop scenes improve visual grounding for computer-use agents

DeskForge generates densely annotated, configurable desktop scenes that improved four vision-language models on grounding benchmarks and multistep computer tasks.

Big Tech
A. Said Gurbuz · Ahmed Nassar · Sunghwan Hong · Marc Pollefeys · Peter W. J. Staar

ETH Zurich · IBM Research Zurich · Microsoft

Research Digest··2 min read
Gurbuz and colleagues built a Linux environment that composes real applications while collecting screenshots, accessibility data, window geometry and action outcomes.

DeskForge controls application state, content, window arrangement, stacking, appearance and display resolution.

Why this paper

From IBM Research Zurich and 2 others

In one line

Controllably composed real desktop scenes provide dense supervision that improves GUI grounding and long-horizon computer-use task completion.

What we could check

  • ·No code link found
  • ·No weights link found
  • ·No dataset link found
  • ·No compute details found
  • ✓Limitations stated by the authors (2 noted)
  • ✓Reports numbers on named benchmarks (5 benchmarks)

Observed from the paper text and links we have. Absence here means we did not find it, not that it does not exist.

§

Research Digest

Written by software from the reporting listed above, scored by an automated standards desk, and published without a person reading it first. If something here is wrong, tell the editor and it will be put right.