In the experimentation world an estimate derived from a mathematical regression model is not always acceptable; this resulted in drafting a dynamic working protocol that allows us to adapt the counting algorithm basing on single cases: physical characteristics of the plant (colour, size, height), field structure, plot sowing scheme.
- Foliar development: through an image-guided segmentation algorithm we can estimate foliar development; these are estimates because the foliar insertion angle is not considered
- Plant counting: through a set of high-resolution photographs and an image-guided classification process we provide counts of a given surface with an error margin of 3% covering both false positives and negatives;
- Plant height: other than with a 3D surface model and a soil model, the exact height mode within a given plot can be calculated
- Plot Quality Report: a mix of the three aforementioned assessment methods complemented by a verification process performed by an operator to remove false positives or integrate false negatives; it allows us to create a detailed, objective and verifiable report relating to the collected informationa