Recall = TP / (TP + FN) Precision = TP / (TP + FP) SOLVE F1 Score = Recall * Precision / (Recall + Precision) F1 = (TP / (TP + FN) * TP / (TP + FP)) / ( TP / (TP + FN) + TP / (TP +FP) ) F1 = ( TP^2 / (TP + FP) + (FP + FN) ) / (TP (TP + FN) / (TP + FP) (TP + FN)) + ( TP (TP + FP) / (TP + FP) (TP + FN) ) F1 = TP^2 / (2 TP^2 + FNTP + FPTP) F1 = TP^2 / (TP^2 ( TP^2 + FN / TP + FP / TP)) F1 = 1 / (TP^2 + FN / TP + FP / TP) Thus, F1 Score is a harmonic mean.
About State of the Art in Artificial Intelligence and Machine Vision