Consider a scenario where you have trained a custom image classification model using Azure AI Vision. The model is intended to classify images of animals into different species. After training, you notice that the model's performance is not meeting the expected accuracy levels. What steps would you take to improve the model's performance, including potential adjustments to the training process and data augmentation techniques?