ggsgumi/multi-class-image-classifier-3-classes

Multi-Class-Image-Classifier (3 Classes)

MUST READ

*All notebooks, saved models and dataset folders must be inside your CWD

Download Dataset & Saved Model

https://www.kaggle.com/fushenggg/3-class-cat-dog-car-dataset

*Contains catdogcar2.hdf5 file to load model (Saves time from training)

Unzip the file and place the 'cat-dog-car' folder into your CWD (current working directory)

Download the ipynb files and place in CWD

Drag the saved model and place in CWD

Running the notebook

1. Check where your CWD is located at

To view your CWD import os os.getcwd()

Mine is 'C:\Users\kansh\Desktop\FYP'

2. Loading of folder from CWD (Changes to make)

Windows uses "\" & Linux uses "//" for directory path

3. Ensure all directory path for train_generators are specified as well

There are TWO WAYS to LOAD the folder from directory path

1. filenames = os.listdir('C:\Users\kansh\Desktop\Entire FYP Folder\FYP Notebooks (Workstation)\cats_dogs_cars\data'

2. filenames = os.listdir('Entire FYP Folder\FYP Notebooks (Workstation)\cats_dogs_cars\data')

Understanding Conv2d & Max Pooling layers

https://towardsdatascience.com/a-comprehensive-guide-to-convolutional-neural-networks-the-eli5-way-3bd2b1164a53