I am confused at finding out the exact F1 score of my YOLOv5 model which underwent training for 150 epochs.

Also, how can I know if the model has done well based on these graphs?

Here are the metrics:

 Class     Images     Labels          P          R     mAP@.5 mAP@.5:.95: 100% 11/11 [00:05<00:00,  2.09it/s]
                 all        347        472       0.91      0.895      0.941      0.746
             class1         347        162      0.923      0.914      0.949      0.718
            class 2         347        161      0.885      0.911      0.942      0.877
            class 3.        347        149       0.92      0.859      0.933      0.641

There is an F1 curve image which automatically gets generated. What is the right F1 score based on the metrics and the image? enter image description here


The F-measure is the weighted harmonic mean of precision (P) and recall (R) of a classifier, taking α=1 (F1 score). It means that both metrics have the same importance. In your graph, the confidence value that optimizes the precision and recall is 0.503, corresponding to the maximum F1 value (0.90). In most cases, a higher confidence value and F1 score are desirable.

  • $\begingroup$ Anything above 0.5 in F1 score is said to be good. I understand that the confidence is around 0.5. So would this graph be called very bad, average or good enough? I am new and I am still learning $\endgroup$ Oct 19 '21 at 14:57
  • $\begingroup$ @EverydayDeveloper, I recommend you this article: towardsdatascience.com/…. $\endgroup$
    – Illustrati
    Oct 20 '21 at 17:49

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.