Using the right algorithm to detect aircrafts and then optimising it

[FireBoyAJ24]

1. Get the data into a tensor
2. Transform the image through horizontalflip, verticalflip and re-size to ensure the images are artificially diverse
3. Make a class that has the initial hyperparameters and parameters. In addition, we should have a forward method to allow the algorithm to predict the label of the image
4. Make a training loop and testing loop def
5. Run the algorithms and download the models for future reference.

[FireBoyAJ24]

Overfitting

1. We can increase the number of images in each of the planes
2. We can add a pre trained model paramters and weights which can identify planes from other objects
3. We could have a learning rate decay to allow the model to reach a convergence

Underfitting

1. Add more layers to the model
2. Have more epochs