Vision papers face 'Instagram moment': prioritizing looks over science hurts reproducibility

taiyasaki · x · 2026-08-28

The author notes that numerous vision research papers cannot be reproduced or built upon due to missing training datasets, code, and weights. Academic publishing is having its "Instagram moment," focusing on cool-looking results rather than the science. The author questions the point of this trend.

Original post →

More from Research

Research channel →