Mara Evaluator

Module to evaluate the inference with respect to a ground truth

This page describes how one can run the BisQue Module named MaraEvaluation

Run Instructions

This module takes in a dataset of images along with two valid G-Object names and generates a confusion matrix that represents the similarity or dissimilarity between ground truth annotations and predicted annotations.

Navigate to Module Page

Login >> Analyze >> Maasai Mara (in Groups Column) >> MaraEvaluation

Expected Inputs

• An image or a dataset of images

  • Click here for a sample dataset of input images

• Ground Truth Annotations

  • String representing the name of a G-Object (Ex: ground_truth_annotations)

• Predicted Annotations

  • String representing the name of a G-Object (Ex: annos_from_AI)

Expected Outputs

• High level stats along with confusion matrix

  • A Table that can be exported as CSV

  • A Consfusion Matrix

Better Visualization of Confusion Matrix: