Results ⊹ ࣪ ˖¶

Every analysis in Traitly produces up to four types of output: two results tables (morphology and color), annotated images, and a session report. This section describes each one.

Results tables¶

Traitly returns measurements as pandas DataFrames, accessible through the results object after running analyze_morphology() and/or analyze_color():

analyzer.results.morphology_results  # morphology DataFrame
analyzer.results.color_results       # color DataFrame

When using analyze_folder() for batch processing, both tables are automatically saved as CSV files in the output folder. analyze_folder() also saves a session_report.txt alongside the CSVs. It includes the Traitly version and its dependencies, the date and time of the analysis, the input folder, and all parameters passed to each step of the pipeline. If any images failed to process, an error_report.txt is also saved listing each failed file and the reason for failure:

Results/
├── morphology_results.csv
├── color_results.csv
├── session_report.txt
└── error_report.txt      <- only if errors occurred

Running Traitly from the CLI produces the same output structure as analyze_folder().

A full description of all available columns is available in the Measurements section.

Annotated images¶

Fruits annotation¶

For every processed image, Traitly saves an annotated version that overlays the detected contours and fruit IDs directly on the original image. These are useful for visually verifying that detection and segmentation worked correctly before drawing conclusions from the data.

Results/
└── image_name_processed.jpg

External analysis annotated image — *External analysis — fruit contour, axes, bounding box, and fruit ID label.*

Internal analysis annotated image — *Internal analysis — fruit and locule contours, internal cavity boundary, axes, bounding box, centroids, and fruit ID with locule count label.*

The annotations vary depending on the analyzer and the steps run:

FruitExternalAnalyzer:

Fruit ID label (id 1)
Fruit contour in yellow
Major axis in green, minor axis in blue
Bounding box in light blue

FruitInternalAnalyzer:

Fruit ID and locule count label (id 5: 4 loc)
Fruit outer contour in green
Internal fruit region boundary in yellow
Locule contours in magenta
Major axis in green, minor axis in blue
Bounding box in light blue
Fruit centroid as a cyan dot
Locule centroids as yellow dots

Color-only analysis

If analyze_morphology() is not called and only analyze_color() is run with FruitInternalAnalyzer, the annotated image will show a simplified version: fruit contour, locule contours, internal cavity boundary, and fruit ID labels – without axes, bounding box, or centroids.

Size reference overlay¶

If a circular size reference is included in the image, the annotation also shows the detection result: a light blue bounding box around the detected reference strip labeled with its YOLO confidence score, each circle outlined in red, and its measured diameter in pixels marked with a blue line. This makes it easy to verify that the reference was detected correctly before trusting the calibrated measurements.

*Size reference box and circles detected*

Color checker overlay¶

If a Macbeth color checker card is detected, the annotation draws a green rectangle over each color patch, marking the exact area used for color extraction from that patch.

*Color checker card and its color patches detected*

Session report¶

At any point after running the analysis, you can save the parameters used in the current session by calling save_parameters():

analyzer.save_parameters(output_path="Results/")

This saves two files:

Results/
├── image_name_parameters.txt
└── image_name_parameters.json

The .txt file records every parameter passed to each step of the pipeline — masking, detection, morphology, color — along with the versions of all dependencies. Useful for sharing, reporting, or documenting your analysis.

The .json file contains the same information in a format designed to be used directly by Traitly. You can pass it to analyze_folder() or the CLI to apply the exact same parameters to a new set of images:

# Python
analyzer.analyze_folder(
    "path/to/folder",
    json_path="Results/image_name_parameters.json"
)

# CLI
traitly --fruit_internal -i images/ --json parameters.json
traitly --fruit_external -i images/ --json parameters.json