Colony grounded SAM2: zero‑shot detection and segmentation of bacterial colonies using foundation models

Bacterial colonies grown on agar plates are used for medical diagnostics, risk assessment, and food safety.•Manual colony quantification is time-consuming and error-prone, motivating automation.•Most AI methods require task-specific, manually labeled datasets, which are costly to produce.•We propose a general purpose zero-shot detection and segmentation pipeline for bacterial colony images, enabling:•Robust colony detection and segmentation•Application across multiple datasets without further training•This allows for AI-assisted analysis on previously infeasible data, further supporting human experts.