@Author: Aman Rana
The datset details are as follows:
- data: This folder contains the following folders
- train: This folder contains the training images. There are 300 RGB images of the original dataset.
- test: This folder contains the testing images. There are 180 RGB images of the original dataset.
- trainLabels.csv: This file cotains labels for all the training images.
The code for this project is contined within the two files:
- run.py: This python file is the starting point of the code. It processes the images and uses a custom generator to create batches of any desired size and calls the model.py file for training the model
- model.py: This file contains code for creation of the Convolutional Neural Network (CNN) model and training based on the training batch passed by run.py file.
Note: Since my laptop is not very powerful and I have installed tensorflow usinf conda (no GPU support via anaconda), I have used a small subset of the original dataset for training. I am uploading an even smaller subset of the original kaggle dataset. The original dataset can be downloaded here