Confusion matrix
Also: classification error matrix
A table comparing predicted classes with actual reference classes.
Counts of true/false positives and negatives support class-specific metrics and expose imbalance hidden by accuracy.
Example Obstacle predictions are tallied against labeled obstacle and clear cases.
Common misconception A confusion matrix is meaningful only with defined labels and representative evaluation data.