I am working in a deep learning project for image classification. I am using Vgg16 to create the model and the dataset has 100 classes. The testing accuracy is 98.9% and loss is 0.1731

And I got the following ROC and Average Precision curves: enter image description here enter image description here

I do not now if the ROC and Average Precision result are good results or there is a problem in the curves because the curves are perfect. Do you know why I got ROC =1? Is there any problem in ROC and average precision?

  • $\begingroup$ Is it on train or test dataset? $\endgroup$
    – keiv.fly
    Commented Oct 31, 2018 at 21:35
  • $\begingroup$ This result on the test dataset $\endgroup$
    – N.IT
    Commented Oct 31, 2018 at 21:56
  • 1
    $\begingroup$ I would recommend to check it manually that the results are that good. Choose one picture run the algorithm and see the result. Otherwise I would just say Wow! This kind of precision I can only dream about. I have never ever seen something that good. $\endgroup$
    – keiv.fly
    Commented Oct 31, 2018 at 22:40
  • $\begingroup$ It's not possible to have it as 1 $\endgroup$
    – Aditya
    Commented Nov 1, 2018 at 2:24

1 Answer 1


Mathematically, it is not possible to have the area under the curve equal to 1 if you have a precision less than 1... If you have a doubt, you should check the confusion matrix and calcuate the scores yourself. You'll know directly if you did a mistake or not

A few other remarks:

  • Accuracy is not precision
  • The loss is of no help here

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

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