Receiver Operator Characteristic (ROC)
We can use an ROC curve to determine a good classification threshold in case we have probability estimates over our scores. I.e. the ROC curve can be used for calibrating our classifier.
The ROC plots the true positive rate (TPR) i.e. the sensitivity on the y-axis, and the false positive rate (FPR) i.e. 1-specificity on the x-axis. Different thresholds have different trade-offs between classifying false positives and true positives, and the ROC curve visualizes these trade-offs.
AUC of the ROC
A good classifier will be more in the top left corner of the ROC plot. This means that curves with a larger area under the curve are better classifiers!
The diagonal expresses random performance and has AUC=0.5
Interpretation: The AUC is the probability that the model ranks a random positive example higher than a random negative example.
The Area Under the Curve (AUC) of the ROC curve is indeed useful for evaluating a binary classifier that outputs probabilities for classes because it can be interpreted as the probability that the binary classifier prefers a random positive instance (label=1) over a randomly selected negative instance (label=0).
However, if we have a class imbalance, the ROC AUC is not very appropriate because, unlike precision and recall, it also takes true negatives (the majority class) into account and can overestimate the classifier's performance as a result.
Free ROC
ROC is used for (binary) classification. The FROC is used for object detection (assuming some hit criterion is defined so that we can speak of true positives, false positives etc.)
For each object detection we can provide a likelihood value, so an FROC curve is applicable.
What's the difference?
The FROC is very similar to ROC, but the difference is on the x-axis. In ROC the x-axis plots lists the false positive rate as 1-specificity. In FROC we instead have the average number of false positives (in terms of wrongly detected objects) per scan.
One major advantage of such a (F)ROC curve is that you can go to medical specialists and ask them: how many false positives per scan is acceptable? This is easier to understand than asking them what f1-score they want.
The "high threshold" will show up on the left side of the FROC, because this will on average result in less false positives.