Regarding a classification problem where for example given an image which depicts a human and we are trying to predict their stance and their behavior. For example Human 1: 'Sitting' and 'Eating' in the first image whilst Human 2: 'Standing up' and 'laughing' in the second. What is the appropriate way of applying the confusion matrix on the predictions. Do I have to unify the predictions?

E.g we have 5 different stances and 5 different behaviors, as a result the confusion matrix is of size 25x25 because we have 25 different classes. Or is there any other way of dealing with such problems?

Is it possible to do the same with multiple objects on an image and how?


1 Answer 1


Confusion matrix is generally not considered as a useful tool to evaluvate our model for multiclass classification and we rather use what is known as AUC-ROC curve.

AUC stands for Area Under Curve whereas ROC stands for Reciever Operating Curve. AUC - ROC curve is a performance measurement for classification problem at various thresholds settings. ROC is a probability curve and AUC represents degree or measure of separability. It tells how much model is capable of distinguishing between classes.

Here is a detailed explaination about the AUC-ROC curve.


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.