Looking at the code "Accuracy", I see that there is an option for the second output / "top" for this layer.What makes this second conclusion?
"Accuracy"
Looking at accuracy_layer.hppwhere the number of exits for a layer is defined, there is this comment:
accuracy_layer.hpp
// If there are two top blobs, then the second blob will contain // accuracies per class.
So, the second "top"layer "Accuracy"simply reports the accuracy of each class.
"top"
, , "" ( , normalization: VALID ).
normalization: VALID
Source: https://habr.com/ru/post/1684593/More articles:How to apply restrictions at all business points in spring / hibernate enterprise application without change codes? - javaUse PrintGCApplicationStoppedTime flag in java 9 - javais it better to use sleep () or an infinite loop to wait for an event? - cWhat is imbalance in image segmentation? - image-processingQueue calls and wait for state updates in React.js - javascriptAssigning test branches in Perl for SUPER :: methods - unit-testingHow to calculate the total for a certain date or better? - sql-serverhttps://translate.googleusercontent.com/translate_c?depth=1&pto=aue&rurl=translate.google.com&sl=ru&sp=nmt4&tl=en&u=https://fooobar.com/questions/1684596/convert-dict-of-lists-of-tuples-to-dataframe&usg=ALkJrhhFzQGrfIPgdGWeO1kfFp7S-wgbWAOutput size Launch problem TensorFlow graph Android - androidJWPlayer + ReactNative + Android: black screen with sound - androidAll Articles