Is it possible to see which instances are classified incorrectly for the trainImageCategoryClassifier example?

Question

Michael Cooper-Stachowsky 2017-8-1

0
链接

此问题的直接链接

https://ww2.mathworks.cn/matlabcentral/answers/351104-is-it-possible-to-see-which-instances-are-classified-incorrectly-for-the-trainimagecategoryclassifie

回答： Michael Cooper-Stachowsky 2017-8-1

I'm getting some good results going through the steps in the trainImageCategoryClassifier example (https://www.mathworks.com/help/vision/ref/trainimagecategoryclassifier.html?s_tid=doc_ta). However, all I can see at the moment is the confusion matrix. I'd like to know which of my testing samples are being misclassified so that I can evaluate them and see if there is anything that is obvious about them that might warrant further study.

0 个评论
显示 -2更早的评论隐藏 -2更早的评论

请先登录，再进行评论。

请先登录，再回答此问题。

Answer 1

Michael Cooper-Stachowsky 2017-8-1

0
链接

此回答的直接链接

https://ww2.mathworks.cn/matlabcentral/answers/351104-is-it-possible-to-see-which-instances-are-classified-incorrectly-for-the-trainimagecategoryclassifie#answer_276329

在 MATLAB Online 中打开

I've answered my own question with a bit of searching, sorry!

%output classifications
for i = 1:length(testSet.Files)
    [idx,score] = predict(categoryClassifier,readimage(testSet,i));
    if (testSet.Labels(i) ~= categoryClassifier.Labels(idx))
        i %just output the index, do interesting things with it later
    end
end