Traitly Architecture ⊹ ࣪ ˖

Traitly is built around a class structure in Python that clearly separates two types of phenotypic analysis: internal and external fruit analysis. This organization lets you choose the level of detail you need without loading functionality you won't use.

The main module traitly.fruit_phenotyping contains two core classes:

# Cross-section analysis (with locules)
from traitly.fruit_phenotyping import FruitInternalAnalyzer

# Whole fruit analysis (external contour only)
from traitly.fruit_phenotyping import FruitExternalAnalyzer

What does each class do?

Class	Focus	What it detects	Typical application
FruitInternalAnalyzer	Internal morphology	Full fruit contour and each individual locule	Cross-sections where the goal is to quantify internal organization (locule count, relative area, symmetry, pericarp thickness, etc.) and the color of different tissues (pericarp and locules)
FruitExternalAnalyzer	External morphology	The outer fruit contour only	Whole fruits for shape, size, and external color studies

Both classes share the same pipeline logic — image processing, segmentation, contour extraction, and trait calculation — but are optimized for their respective goals:

• FruitInternalAnalyzer looks for hierarchical relationships: a fruit contour containing multiple locule contours within it.
• FruitExternalAnalyzer focuses on the full fruit silhouette, ignoring internal structures.

General workflow for each analysis. A) FruitExternalAnalyzer: pipeline for whole fruit external appearance analysis. B) FruitInternalAnalyzer: extended pipeline for fruit and locule detection and segmentation in cross-sections.

---

How to use Traitly?

Traitly is available in three environments:

From Python (Jupyter Notebook or script): recommended for interactively exploring and adjusting parameters on individual images, and for efficient processing of large image batches. If you don't have Jupyter installed, refer to the official documentation.

From the terminal (CLI): to run the analysis directly from the terminal without writing a Python script. This is especially useful on servers or computing environments without a graphical interface. See the CLI section for details.

From the Shiny app: for interactive analysis without writing code. Available locally by running traitly-app in the terminal (see the Shiny App section for more details), or online through the interactive demo.

The following sections contain detailed documentation for each environment, including usage examples, configurable parameters, and extracted traits. See also the image specifications for internal analysis and external analysis to ensure the best results.