Skip to the card

Card 456 of 9982026-09-01 issue

Observed arrival · 2026-09-01

Label Refinery’s self-labeling construction-site loop

labelrefinery.org Observed source
Editorial interest 84/100 Selection signal · not a rating of the site

A technically explicit perception project that uses geometry, learned models, and synthetic construction scenes to generate and improve its own training labels.

Landing page captured for the 2026-09-01 issue.

Field notes

The documented loop preserves each pipeline result as an immutable dataset version, allowing planning to read what the previous round recorded rather than overwrite it. Its first stage uses LiDAR sweeps and joint states to discover unnamed instances; later stages add class names and retrain a detector. The evaluation setup uses a synthetic 60-second construction scene with a held-out oracle, while Grounding DINO supplies names only after geometry has proposed the objects.

Observed signals

Read the marks

Editorial observations of this landing page, not a rating.

OpenPublic substance visible
PrettyNotable craft visible
ProPolished or operationally mature
NicheUnusually specific use

One card from the complete issue

Nobody Gets to See the Answer

315,616 arrived 1,000 judged 998 catalogued Enter the complete issue
labelrefinery.org

Landing page observed 2026-09-01. The live site may have changed.