Classification of Dog and Cat Vs Classification of different breeds of dogs

Question

Case1: Say I train a classification model to classify an image whether it's a dog or a cat.

Case2: Also I need to train a classification model to classify different breeds of dogs, say 5 different breeds.

In case1, the model can easily learn the pattern to differentiate between dog and cat.

But when solving Case2 should I do something different, because I need to classify among different breeds of dogs, which can be a little difficult for the model to the learn.

Answer 1

The approach for case 2 should be approximately the same as case 1. The only changes will be:

• You may need more training data to reach acceptable performance due to the harder nature of the task.
• The neural network architecture will be different as i) it is no longer a binary classification problem and ii) the transformations which need to be learned are more complex.