Traitly

Traitly is a Python-based tool designed to automate the analysis of fruit images, supporting both single-sample and high-throughput workflows. It addresses the need for reproducible and unbiased extraction of morphological, structural, and color traits from standard RGB images of fruits, including both internal (cross-section) and external (surface) views. The platform is intended for scientific research, particularly in plant breeding and genetics, where large-scale phenotyping is common.

Key features of Traitly include the automated extraction of traits such as color, shape, and size from fruit images. For internal images, it can measure characteristics like the number and distribution of locules, pericarp thickness, symmetry, and internal tissue color. For external images, it assesses general fruit shape, size, and surface color. Traitly supports both single image and batch processing, assigning a unique ID to each detected fruit and generating per-fruit measurements in the output. The tool is fully automated, handling detection, segmentation, calibration, and trait extraction without manual intervention, which reduces bias and saves time in phenotyping studies.

Traitly incorporates pre-trained models for tasks such as automatic detection of size reference markers and sample labels, and it offers color correction using Macbeth Color Checker detection to standardize results across experiments. It can also identify QR codes and text labels for sample identification. The platform supports direct conversion of scanned PDF files to images and automatically generates session reports that record parameters, dependency versions, and metadata for each analysis, ensuring traceability and reproducibility.

The tool is available for use in multiple environments, including Jupyter Notebook, command-line interface (CLI), and a web application built with Shiny. An interactive online demo is also provided, allowing users to try Traitly without local installation. Documentation is accessible in both English and Spanish. Traitly is built on established Python scientific libraries such as OpenCV, NumPy, SciPy, pandas, and matplotlib, enabling high performance even in large-scale analyses. Its modular architecture supports adaptation to other biological tissues, such as seeds or leaves, with minimal adjustments.

Traitly is positioned as a collaborative and extensible project, encouraging contributions from the scientific community to expand its capabilities and documentation.