Segment Any Plant (SAP): Foundation-Model Segmentation for Plant Time-Series Phenotyping.

SAP enables training-free, high-accuracy segmentation of plant time-series images across species and developmental stages, achieving mean IoU of 0.89–0.93 for macroscopic organs.

• Why it matters: Automated plant segmentation is essential for quantitative growth and environmental response studies but is hindered by the need for extensive retraining and annotated datasets, limiting cross-species and condition applicability.
• What they did: The framework leverages the pretrained Segment Anything Model 2 (SAM2) with interactive prompts, automated mask propagation, and centerline extraction, requiring no additional training and supporting diverse plant imaging systems.
• The result: SAP facilitates reproducible, scalable phenotyping of plant growth and morphology, enabling detailed analysis of organ dynamics and opening avenues for broader application in plant science research.