Blog
How SASYA Works
Notes on mapping real-world plant identification skills — leaf morphology, taxonomy, dichotomous keys — into an on-device app. Written by the people building it, warts and all.
- August 1, 2026Identifying Plant Families by Pattern: Turning a Printed Key into DataMost plant apps answer 'what is this?'. Family-level pattern recognition answers 'how would I know?'. Here is how we encoded the patterns method as a trait ontology, what it can resolve, and where the data runs out.
- July 7, 2026Choosing a Model for Plant Disease Detection: Datasets, Architectures, and TrainingA practical guide to plant disease detection: how to actually use a dataset like PlantVillage, how to choose a model architecture for your use case, and why fine-tuning an existing model usually beats training one from scratch.
- July 5, 2026How SASYA Identifies a Plant — On Your Phone, OfflineAn engineering deep-dive into SASYA's two-stage vision pipeline: a real-time detector that finds the plant, a 3,559-way classifier that names it, and the context re-ranking and learned embeddings that make a single photo more reliable.
- July 5, 2026Encoding Leaf Morphology as a Feature SchemaHow SASYA turns the botanical vocabulary of leaf shape, margin, and venation into a structured, filterable trait schema — and where that data actually comes from.