Figure 2. Diagram illustrating calculation of each metric used in training (train and test) data: Precision, Recall, Accuracy and F1 (range of 0 to 1). For identification purposes, misidentifications are counted as correct (green in confusion matrix) because the animal was detected; whereas, for classification purposes, misidentifications are counted as incorrect (red in confusion matrix) because the object was not classified correctly. True Positives (TP), False Positives (FP), and False Negatives (FN) are represented in the confusion matrix with True Negatives (TN) not present in training data. Adjusting confidence thresholds (range of 0.5 to 0.95) optimizes the model for specific applications.