Load classifications from CSV file (validation mode)

Reads a classification CSV file and returns a data frame with classifications. Class names are processed to truncate trailing numbers (matching iRfcb behavior).

Usage

load_from_csv(csv_path, use_threshold = TRUE)

Arguments

csv_path

Path to classification CSV file

use_threshold

Logical, whether to use the threshold-filtered class_name column (default TRUE) or the raw class_name_auto column when available.

Value

Data frame with classifications. Expected columns: `file_name`, `class_name`, and optionally `score`.

Details

The CSV file must contain the following columns:

file_name

Image filename including the `.png` extension (e.g., `D20230101T120000_IFCB134_00001.png`).

class_name

Predicted class name (e.g., `Diatom`).

Optional columns may also be included:

score

Classification confidence value between 0 and 1.

class_name_auto

Raw (unthresholded) class prediction. When use_threshold = FALSE and this column exists, its values are used as class_name.

The CSV file must be named after the sample it describes (e.g., `D20230101T120000_IFCB134.csv`) and placed inside the Classification Folder configured in the app (subfolders are searched recursively).

Examples

if (FALSE) { # \dontrun{
# Load classifications from a CSV file
classifications <- load_from_csv("/path/to/D20230101T120000_IFCB134.csv")
head(classifications)
} # }