How to optimize the existing approach

As you have seen in the previous section, because of the lack of computation hardware, we have achieved a 66% accuracy rate. In order to improve the accuracy further, we can use the pre-trained model, which will be more convenient.

Understanding the process for optimization

There are a few problems that I have described in the previous sections. We can add more layers to our CNN, but that will become more computationally expensive, so we are not going to do that. We have sampled our dataset well, so we do not need to worry about that.

As part of the optimization process, we will be using the pre-trained model that is trained by using the keras library. This model uses many layers of CNNs. It will be trained on multiple GPUs. So, we will be using this pre-trained model, and checking how this will turn out.

In the upcoming section, we will be implementing the code that can use the pre-trained model.