CNN (Convolutional Neural Network) Matlab or Python

I want to classify the images into eight labels using CNN (convolutional neural network) or other deep learning techniques (GAN etc). These images are the vessels (merchant ship and towing ship) light at night. I cut the noise from the raw image using Paint program by manual. Challenge; 1) The pixel size of all images is different. It is because of cutting the image using the Paint program by manual. I think all images need to be rescaled. 2) As you can see, the data is not enough. It needs to be augmented using rotating, noise etc. Requirement; 1) augmentation; rotating angle [-10, 10], noise(salt and pepper etc) 2) divide the images into train and test data 3) model (layer); VGGNet, GoogleNet, ResNet, enseNet or reconstruct the model 4) drop out to reduce overfitting 5) Loss(cross entropy) 6) precision and recall Please make a code for this project using Matlab or Python. I attached the powerpoint for understanding this project well. I will share all images by Google Drive. [login to view URL]