Fish Fry Growth Rate Measurement via Reference Object

Problem to Solve

In fry farming, growth rate is a key indicator for health and feeding suitability, but manual daily length measurement is time-consuming, and ordinary images do not directly provide millimeter-level data.

How the Skill Works

The skill analyzes still images or video from a fixed fry-tank camera. The view must include a known-size reference object such as a scale ruler, standard coin, or calibration board. Inputs can be local files or network URLs, supporting jpg/png/mp4/avi/mov up to 10MB, with a recommended resolution of at least 1080p. The main workflow is:
- Detect the reference object and convert pixel length to actual mm
- Measure fry body length from snout to tail-fin tip
- Estimate growth rate in mm/day, population mean, standard deviation, and CV%
- Generate growth curves, risk alerts, and suggestions by tank_id and timestamp

It also uses species, age, and water temperature to choose baselines for zebrafish, tilapia, koi fry, and angelfish, rather than applying a single generic threshold. If growth is unreliable, stagnant, or uneven, it returns staged alerts.

Boundaries

The camera should shoot vertically from above, and the reference object should be on the same plane as the fry. If reference-object confidence is low, fish posture is severely bent, or the view is heavily occluded, the result should be marked growth_measurement_unreliable. It does not diagnose specific diseases, prescribe drugs or dosages, or recommend specific feed brands. Historical report queries are read only from the cloud API and rendered as Markdown tables.