ComputerVisionMetrics

ComputerVisionMetrics.jl is a Julia package for evaluating segmentation models using various metrics like Dice and Hausdorff distances. The primary functions are dice_metric() and hausdorff_metric() which evaluate the similarity and dissimilarity between predicted and ground truth segmentations respectively.

• Supports 2D and 3D segmentation masks
• Provides functions to calculate Dice and Hausdorff metrics
• Offers additional utility functions like get_mask_edges and euc for edge detection and Euclidean distance calculation

using Pkg
Pkg.add("ComputerVisionMetrics")

Basic usage involves passing your predicted and ground truth segmentation masks to the metric functions:

using ComputerVisionMetrics

# Load your data
prediction = rand([0, 1], 10, 10, 10)
ground_truth = rand([0, 1], 10, 10, 10)

# Evaluate Dice metric
dice_score = dice_metric(prediction, ground_truth)

# Evaluate Hausdorff metric
hausdorff_score = hausdorff_metric(prediction, ground_truth)

For more advanced usage and options, see the documentation