Biosimulant

What should we check before trusting an automated colony count from a plate photo?

demiBiosimulant team · 1 replyAnswered by agent

Starter question from the Biosimulant team.

For automated counting of visible colonies, what should a person check in the annotated image before using the reported count?

Could we use a public example to discuss merged colonies, missed detections, image conditions and agreement with manual counts? It would also help to explain the difference between counting visible colonies and calculating CFU/mL, and how to decide whether a counter is suitable for a new laboratory's images.

An experimental example is the Plate Colony Counter, release demi/plate-colony-counter@0.1.1. Human review is required; this release did not pass its original per-plate count-error gate.

1 reply

  1. Biosimulant Agent AI agent · official

    Here is what Biosimulant's own guidance says about this:

    1. Plate colony counter - YOLO 0.1.1
    2. Plate colony counter - YOLO 0.1.0
    3. Package and publish with Biosimulant: Publishing cuts an immutable namespace/name@version release on the Hub; a change means a new version, never an edit.

    If this does not answer your question, mark the post as needing a human and a person from the community or the Biosimulant team will follow up.

    Written by an AI agent. Check its sources before you rely on it.