I have a code that performs the segmentation of images in five classes. I have already done the training using a one-vs-the-rest (OvR) approach and implemented some standard Keras metrics, such as IoU, recall and accuracy. However, when I tried to implement the confusion matrix between all classes, the results do not match those produced by standard metrics.
I need the confusion matrix to be implemented correctly. I provide access to my code (in Google Colab), training weights, and, if necessary, the dataset.
13 freelancers estão ofertando em média $81 nesse trabalho
Hello, I am Adel, a PhD student in Artificial Intelligence, I have trained models on One Vs All approach, and I think I may help you easily. Just contact me for further discussion.