We saw how exceptionally well deep learning models performed when applied to computer vision and classification tasks. Our LeNet model with the MNIST and Fashion-MNIST datasets was able to achieve 90%–99% accuracy under a very reasonable amount of training time. We have also seen how the ImageNet models have achieved record-breaking accuracy levels in more complex datasets.
Now you might be eager to try out what we learned on a more complex and practical classification task. But what should we consider when we are going to train our own image classification model with our own categories ...
